Every market map classifies AI companies by sector: infrastructure, enterprise, vertical, health, climate. The Headline AI Europe 100 classifies companies according to what they sell and what they replace. This test sorts an AI company in seconds and predicts business models, metrics, and moats, unlike other market maps.

The Europe 100’s six breeds are deliberately heterogeneous. Take a voice model, a sovereign LLM developer, and a biology foundation model: a sector map files them under media, enterprise, and health – three columns with seemingly nothing in common. This map puts them into one breed because they sell intelligence – the model itself – and are analyzed using a similar framework. What a company sells, what it replaces, how capital converts into moat, where margin compounds, and where it is capped: these shared patterns determine how the analysis should be done.

This year, we decided to release our framework: each category has a corresponding animal that exemplifies its unique qualities.

Reading note. The framework is global, and the lens is European, with European players marked [EU] throughout. Key figures were re-verified against public sources on July 3, 2026.

The six breeds at a glance

How to classify a company in 10 seconds

One question per breed. Ask them in order; stop at the first yes.

01 · Frontier Labs

Foundational models · The Whale

Every other breed buys from, builds on, or defends against this one. Frontier Labs sell cognition itself, priced by the token, and the leaderboard resets with every release.

Overview

These companies are building the intelligence layer. The product improves in discrete jumps, each jump resets the competitive order, and the cost of staying in the race grows every round.

One breed, several frontiers

Foundation model does not mean language model. This breed is defined by selling the model itself, not by what the model processes. But each modality prices differently.

• Text and reasoning. The largest TAM, the heaviest capital. OpenAI, Anthropic, Mistral, DeepSeek.

• Image and video. The moat comes from distribution and consistent output. Benchmark scores matter less. Black Forest Labs.

• Voice. The first modality settled by enterprise distribution rather than model quality. ElevenLabs.

• World models. Models that predict the physical world: pre-revenue by design, priced on team and thesis. AMI Labs.

• Scientific and tabular data. Biology, chemistry, tabular. Narrower TAM, deeper defensibility: proprietary data replaces scale as the barrier. Bioptimus. Fundamental.

The modality changes the diligence. The KPI does not.

Market dynamics

• Winner-takes-most, never for long: the lead reshuffles every 3 to 12 months on release cycles.

• States sit inside the demand curve. Sovereignty requirements, export controls, and national champions decide who buys, who sells, and at what price.

Business model

• API revenue priced per token; margin tracks the inference cost curve.

• Consumer subscriptions, ChatGPT, Claude, Le Chat: the second engine, and where the brand is built.

• Sovereign and enterprise contracts: multi-year, custom, lumpy, and the hardest to displace.

Distribution

• The broadest TAM on the map: developers, then prosumers, then enterprises and states.

• Two clocks run at once: API adoption in seconds, sovereign deals in quarters.

• ACV from cents per call to nine figures, inside one P&L.

Metrics

• Revenue scaling 5 to 10x annually for category leaders.

• Developers and prosumers defect on every rival release; enterprises stay once integrated.

• Gross margins negative to thin. The breed is a bet that inference costs fall faster than prices.

Moats

• Pre-revenue, the moat is talent: the labs that concentrate the best researchers get to exist.

• Post-revenue, it shifts to compute access and distribution.

• Capital is not a metric or a moat – none is defended by it.

Investor lens

• Rounds are priced in billions. Getting access to one is the first constraint, ahead of judging whether to write the check.

• Pre-revenue signal: the net flow of senior researchers, who leaves what to join whom.

• Exits split in two: IPO for the winners, absorption for the rest. Sovereign consolidation is already the European pattern.

• Risks: open weights closing the gap, model-layer commoditization, the reshuffle, unproven profitability.

Key examples

• OpenAI: category-definer, agents, and consumer distribution.

• Anthropic: enterprise traction.

• Google DeepMind: vertical integration with Google distribution.

• Mistral AI [EU]: European sovereign play.

• ElevenLabs: voice.

• Black Forest Labs [EU]: image generation leader.

• Fundamental Technologies [EU]: tabular model.

• AMI Labs [EU]: world models.

• Bioptimus [EU]: foundation model for biology.

• Recursive Intelligence [EU]: world models.

• Ineffable [EU]: RL foundational model.

• Stability AI [EU]: image and video models.

• DeepSeek: Chinese, challenged the scaling orthodoxy.

02 · Velocity Racers

AI-native applications · The Cheetah

The fastest revenue curves in software history live here. So do the fastest collapses. Velocity Racers package AI into products that a user can adopt in minutes, though they can abandon those products just as quickly.

Overview

Self-serve, AI-native applications. The moat is rarely the model. It is iteration speed, product taste, distribution, and owning the value proposition in the user’s head.

Market dynamics

• A category leader emerges in under six months, often on a single product moment.

• Markets fragment fast: one clear leader, a long tail of replicas.

• The US leads on volume; Europe wins selected niches.

Business model

• Freemium and prosumer subscriptions, $10 to $50 per user per month.

• Usage add-ons, credits and generations, as the expansion lever.

• A light enterprise tier as the upsell path. Low ACV by default.

Distribution

• ICP: consumers, prosumers, SMBs, individual professionals.

• Sales cycle in minutes: self-serve, bottom-up.

• ACV of $100 to $5K per account; virality compensates.

Metrics

• $0 to $10M ARR in six to 12 months, $100M in 24, for the winners.

• Retention is the structural weak spot: the wow-effect fades, novelty churn is real.

• Burn multiples run high. Marketing spend buys mindshare, and mindshare is the asset.

Moats

• Velocity itself: everyone ships weekly now, so cadence alone stopped being a real edge.

• Brand: being the name the ICP associates with the job to be done.

• Taste over technical depth: the founding team’s design sense matters more than the model choice.

Investor lens

• Capital funds marketing and the perception of being the fundraising leader; both are self-reinforcing.

• Exit: IPO for category leaders, acquisition for the rest.

• Risks: feature-not-company, foundation model dependency, OS-level distribution capture.

Key examples

• Lovable [EU]: the archetype; $0 to $100M ARR in eight months.

• Cursor: AI code editor, $60B agreed sale to SpaceX (Jun 2026).

• DeepL [EU]: translation, the quiet compounder of the breed.

• Synthesia [EU]: AI video for enterprise.

• Granola [EU]: meeting notes, prosumer wedge.

• Higgsfield: AI video and creative generation.

• Perplexity: consumer search repositioning toward browser and enterprise.

• Midjourney: AI image generator, early category-definer, still independent.

03 · System Builders

Deep AI platforms · The Octopus

The next ERPs, CRMs, and operating systems are being written now. System Builders replace the software an enterprise runs on.

Overview

The playbook is land-and-expand: enter through a wedge, spread into the platform. Harvey from document review, Glean from search, Legora from drafting. Once the workflows, data, and habits move in, it’s difficult to pull these systems out.

Market dynamics

• Every legacy category is under attack: ATS, CRM, EHR, legal tech, ERP modules.

• The window is set by the incumbents: entrench before Salesforce, SAP, and Microsoft ship good-enough agents into their installed bases.

• Compliance shapes procurement: SOC2, HIPAA, ISO, FedRAMP are the tickets to the table in regulated verticals.

Business model

• Per-seat subscriptions: the customer pays for access to the tool, regardless of how much work it actually does.

• Tiered platform pricing: modules, add-ons, and premium support drive expansion.

• Consumption and outcome pricing are eroding the per-seat default from inside the breed.

Distribution

• ICP: mid-market and large enterprises, regulated verticals first.

• Sales cycle of six to 12 months, multi-stakeholder.

• ACV from $100K to several million, expanding with seats and modules.

Metrics

• Growth is slow, then violent: Harvey went from $50M to $190M ARR in 12 months.

• NDR of 130%+ marks the best, and drives most of the long-term value.

• Gross retention of 95%+ once embedded. What makes switching costly is how deeply the product sits in daily operations; the contract terms barely factor in.

Moats

• Workflow embedding: data, integrations, and user habits compound into stickiness.

• Compliance and trust: earned barriers new entrants cannot shortcut.

• A double talent bar: enterprise sales DNA and AI research credibility, both required, rarely found together.

Investor lens

• Capital-hungry: the enterprise sales ramp is slow and expensive before it compounds.

• Exit: IPO for category leaders, strategic M&A by enterprise incumbents.

• Risks: incumbents shipping good-enough AI, and implementation drag that stalls the expand motion.

Key examples

• Harvey: legal, category-definer.

• Legora [EU]: legal, contender.

• Glean: enterprise search and work assistant.

• Attio [EU]: new-generation CRM.

• Pigment [EU]: France, AI-native planning.

• n8n [EU]: workflow automation.

• Sana [EU]: learning and knowledge.

• Serval [EU]: AI-native IT operations.

• Payflows [EU]: new-generation ERP.

• Aikido [EU]: security platform.

• Abridge: healthcare documentation.

• Pelico [EU]: AI-enabled manufacturing operations.

04 · Outcome Engines

Agentic services · The Beaver

The customer does not buy software here. It retires a cost line. Outcome Engines absorb a function and bill for the result.

Overview

The thesis is all about labor compression: AI shrinks the cost of delivering a result faster and spreads throughout the business. If the compression stalls, what remains is a services company at services multiples.

Market dynamics

• The targets are labor-intensive functions: customer service, claims, sales development, recruiting.

• The legacy enemy is the services incumbent, where a 10x cost reduction is on the table.

• ROI is measured against an existing budget, so the sale closes faster than any software deal.

Business model

• Per-outcome pricing: per resolution, per claim processed.

• Performance contracts: SLAs on volume, quality, and response time.

• Hybrids emerging: retainer plus a per-outcome premium tier.

Distribution

• ICP: any company running a labor-heavy function, mid-market to large enterprise.

• Sales cycle of two to six months, shorter than System Builders, because the ROI is arithmetic.

• ACV scales with the volume of work absorbed: $50K to multi-million.

Metrics

• Growth follows the proven SLA; account expansion follows the automation rate.

• Gross margin starts at 30 to 50% and climbs toward 70 to 85% as automation rises.

• The automation curve is the whole story. Flat curve, services multiple.

Moats

• Operational excellence: delivering the outcome reliably at scale is harder than building the model.

• Data flywheel: every interaction trains the system that handles the next one.

• Domain depth in the target vertical, which generalist AI cannot cross quickly.

Investor lens

• Moderate capital: revenue can fund growth when the model holds.

• Exit: M&A by service incumbents, IPO if margins prove software-like.

• Risks: SLA failure is binary, and if the automation rate stalls, the company gets re-rated as a services business rather than just slowing down.

Key examples

• Sierra: customer service, category-definer.

• Decagon: customer service.

• Parloa [EU]: AI contact center.

• Mercor: recruitment and talent matching.

• Notch: complex support for regulated industries.

• EvenUp: legal claims, performance-based.

05 · Stack Enablers

AI infrastructure · The Bee

Stack Enablers sell to all players and grow with the sum.

Overview

Three sub-layers, three economics: compute and serving, data and training, dev tooling and orchestration. What they share is the position: everyone else’s growth is their revenue.

Market dynamics

• Revenue indexes on total AI consumption, which keeps compounding regardless of who wins above.

• The prize is default status: becoming the standard in a developer workflow beats any single contract.

• The squeeze comes from both directions: labs integrate down, tooling consolidates up.

Business model

• Usage-based dominant: tokens routed, compute provisioned, experiments run.

• Capacity reservations: long-term contracts locking GPU supply for labs and hyperscalers.

• Freemium and open source for dev tools; production usage and team features convert.

Distribution

• ICP: developers, labs, hyperscalers, enterprises deploying AI internally.

• Minutes for a dev tool, 4 to 12 months for an enterprise contract.

• ACV from a few dollars to nine-figure deals.

Metrics

• Consumption growth per account, compounding with total token demand.

• Gross margin moderate on compute, mixed on data, best in tooling.

• Retention is high once embedded: churning means running a migration project.

Moats

• Integration depth: unwinding it means staffing an engineering project. A renewal call won’t do it.

• Developer ecosystem: community, standards, and default-tool status.

Investor lens

• Capital: billions for compute, moderate for data and tooling.

• Exit precedent set by CoreWeave’s IPO (Mar 2025); M&A by hyperscalers and data platforms for the rest.

• Risks: vertical integration from above and below, and demand concentrated on a handful of labs that can become competitors.

Key examples

• CoreWeave: GPU cloud, IPO Mar 2025, acquired Weights & Biases (May 2025).

• OpenRouter: model routing and aggregation.

• Together AI and Fireworks AI: open-model inference.

• Hugging Face: the standard model repository.

• Nscale [EU]: AI infrastructure.

• Fluidstack: AI compute.

• Baseten: model deployment and serving.

• Runware: media-generation inference.

• Dataiku: data science and AI platform.

• Dash0 [EU]: observability.

• Langfuse [EU]: open-source observability.

06 · Intelligent Hardware

Physical AI · The Rhino

Intelligent Hardware embeds autonomy into machines and sells it through the slowest, most protected procurement on earth. That slowness is the moat.

Overview

Robotics, autonomous defense, space, and dual-use hardware. Capex-heavy, regulation-bound, and geopolitical by construction.

Market dynamics

• Sovereign demand is the engine: national security budgets drive the largest contracts.

• Development cycles run three to 10 years from concept to deployment.

• ITAR, export controls, and EU defense rules define what can be built and sold, and to whom.

Business model

• High-unit-cost hardware sales: drones, robots, satellites, ground systems.

• A software overlay, autonomy stack, data services, fleet management, adds the recurring layer.

• Multi-year government programs with milestone-based payments.

Distribution

• ICP: defense ministries, prime contractors, space agencies, industrial conglomerates.

• Sales cycle of 12 to 36 months, RFP-driven, longer for defense and space.

• ACV in the millions to hundreds of millions; nine-figure single contracts happen.

Metrics

• Growth is lumpy and tied to individual contracts rather than a steady monthly recurring line.

• Gross margin swings with the hardware-to-software mix.

• Backlog and program wins lead revenue. ARR is the wrong lens in this breed.

Moats

• Manufacturing, supply chain, and certification: years to replicate, impossible to shortcut.

• Security clearances and trusted-vendor status act as structural barriers; no amount of competing on price gets around them.

• Autonomy tuned to the specific physical system, hard to copy across platforms.

Investor lens

• Capital: hundreds of millions to billions across the lifecycle.

• Exit: IPO for the largest; M&A by primes for the rest.

• Risks: program cancellation, geopolitical exposure, capex overruns, certification delays.

Key examples

• Anduril: autonomous defense, category-definer.

• Helsing [EU]: defense AI, category-definer.

• Harmattan AI [EU]: autonomous defense drones.

• Quantum Systems [EU]: dual-use drones.

• Stark [EU]: autonomous defense systems.

• Wayve [EU]: end-to-end autonomous driving.

• RobCo [EU]: industrial robotics.

• Wandercraft [EU]: exoskeletons and the Calvin industrial humanoid.

• Skild AI: foundation models for robotics.

• Figure AI: humanoid robotics.

• SpaceX: launch and orbital infrastructure.

Reading the map

From taxonomy to thesis: where value is created, threatened, and mispriced.

The six breeds are a sorting tool. What follows is the thesis: the memory system that makes the grid stick, where capital buys defensibility and where it cannot, how margins travel as each breed scales, which migrations between families create or destroy value, and what the map says about Europe.

The Animals

The animals are a memory system: each one chosen because its actual biology encodes the breed’s economics. Remember the animal, and you remember what to diligence.

01 · The Whale. The largest creature in the ocean feeds by filtering colossal volumes of the smallest target. Its whole economic problem is turning that intake into mass, and that’s the animal problem too, just with computing power instead of plankton. Whales are few, they’re tracked by name, and every time one surfaces it makes the news, much like a model release does. A whale can’t survive in a small habitat; it needs an ocean, and a frontier lab needs an entire capital ecosystem to match.

02 · The Cheetah. Zero to a hundred in three seconds. This breed runs zero to $100M ARR on roughly the same timeline. Everything in the cheetah’s body is built for acceleration and nothing is spare: taste over technical depth rendered in anatomy. It is also the most photogenic predator on the savanna, the one everyone wants on camera, which is the moat itself: brand, mindshare, being the animal people think of first. But top speed holds thirty seconds, and after the sprint the cheetah rests while lions steal the kill. Its weakness is endurance. The breed’s weakness is retention: you can win the hunt and still lose the meal.

03 · The Octopus. Eight arms, eight workflows: land with a wedge, then expand limb by limb. Net Dollar Retention (NDR) is essentially a count of how many arms are still gripping, and once the suckers attach, removal turns into surgery. Two-thirds of an octopus’s neurons live in its arms rather than its head, so intelligence sits distributed across every module. It also camouflages itself to whatever environment it’s in, much as a platform deeply integrated into its clients’ architecture.

04 · The Beaver. The only animal besides man that engineers its own habitat. What you’re paying for is the dam, never the hours the beaver spends on it: the pond is the outcome, the labor stays invisible. A dam holds water or it fails, and that binary is close to what an SLA measures. And every dam the beaver builds teaches it to build the next one faster: experience compounds, which is the data flywheel in one image. Each season the same dam takes less labor to maintain, and that is the automation curve, playing out one pond at a time.

05 · The Bee. A single bee is unremarkable. The colony is everything. Each bee works the entire field instead of betting on one flower, which is exactly how a Stack Enabler behaves: it captures a sliver of consumption from every AI infrastructure and compounds. Bees pollinate crops that they will never own. And a hive defends itself through structure and standards, just as integration depth and default status make switching feel like relocating an entire colony.

06 · The Rhino. It wins through mass, armor, and momentum, never agility. The armor is the moat: certifications, government contracts, manufacturing capacity, all built up exactly where threats land. Rhinos are among the most geopolitically protected animals on earth, guarded by treaties, armed escorts, sovereign custody, and that’s roughly what it feels like to operate inside ITAR and export controls. They’re territorial too: a program win is a waterhole, claimed and defended.

Capital × Moat

Where capital deepens the moat, and where it cannot.

Plot the six breeds on two axes: capital required on the horizontal, moat defensibility today on the vertical.

One definition first, because the axis fails without it. Capital required is the structural entry barrier: what it costs to exist in the breed at all. It is not burn, which is discretionary go-to-market spend. Velocity Racers burn enormously and require little.

Read the board corner by corner:

• Frontier Labs, top right. The most capital spent and the highest perceived moat on the board, and the most contested position on it. On our read, the model layer is commoditizing: open weights compress the frontier premium, and each release cycle gives challengers a fresh shot. The moat is real today and rented by the quarter.

• Intelligent Hardware, just below. The one breed where capital reliably converts into defensibility, because it buys things competitors cannot shortcut: certifications, program access, manufacturing, clearances. Slower to win, harder to displace. The most durable moat on the board.

• System Builders, Outcome Engines, Stack Enablers, the middle. Moats here are earned, not bought: embedded workflows, data flywheels, integration depth. Capital accelerates the earning; it cannot substitute for it. A competitor with twice the funding and none of the embedding is not ahead.

• Velocity Racers, bottom left. Modest capital required, weak structural moat. This isn’t a flaw; it’s the breed design. Spending more money doesn’t get a company out of this corner. Migrating to a different breed does, and that’s what the next section maps.

Top right is capital-fueled moats, top left is earned moats, bottom left is structural fragility. Bottom right, high capital and low defensibility, is the moat mirage: the most expensive mistake on the board, and where a Frontier Lab lands if commoditization outruns its revenue conversion. The reading rule for an investor: price the moat trajectory, never the capital deployed. Capital tells you who could afford to enter. The moat tells you who gets to stay.

Margin trajectories

Where margin compounds the most, and where it is capped.

Gross margin here means margin after inference, compute, and human-in-the-loop costs, including training amortization and depreciation. Two numbers per breed, today and expectations at scale, and two forces per breed: what pushes margin up, and what caps it. These are rough estimates to give orders of magnitude.

Three readings matter more than the levels.

First, the steepest slopes belong to the two families where the up-scenario is least guaranteed. Frontier Labs travel from a negative margin to 45% only if inference optimization outruns the capex race, and so far, the race is winning. Velocity Racers reach 75% only if pricing holds while usage explodes, and the breed’s own history says users consume faster than costs fall.

Second, System Builders start where everyone else hopes to arrive. 70% today, 80% at scale: software economics from day one. The threat is internal, agentic features that consume tokens faster than they add price. The System Builders margin question is not whether the margin can increase, it is whether the new capabilities pay their own way.

Third, two breeds have ceilings built into their physics. Outcome Engines cap where their automation rate caps: the residual human in the loop is a permanent cost of goods. Intelligent Hardware caps because every unit ships with atoms in it. Neither ceiling is a flaw; both are prices of admission to markets software alone cannot reach. The mistake is paying software multiples for capped-margin businesses, which brings us to migrations.

The migration map

The map is not static. Breeds build defensibility by migrating product or pricing.

A breed describes a company’s current state. It isn’t a fixed identity. Companies are classified by their dominant revenue engine, and that engine can change. When it does, the valuation re-rates on how credible the migration looks, often well before the migration is finished. So the question that matters most in diligence isn’t which box a company sits in today. It’s which arrow it’s moving along.

The value paths

02 → 03. The golden path. Enter with Velocity Racer economics, self-serve adoption, and explosive top line, exit with System Builder economics, enterprise NDR, and platform multiples. It cures the breed one structural disease, novelty churn, by replacing it with workflow embedding. Cursor is the documented case: a product that entered the arrow on prosumer subscriptions exited with enterprise accounting for roughly $2.6B of its ~$4B annualized revenue by June 2026. It will not end the journey independently: SpaceX agreed in June 2026 to acquire it for $60B in stock, the largest acquisition of a venture-backed startup on record, expected to close in Q3. Perplexity is attempting the same migration, in full view of the incumbents it threatens, moving from consumer answer engine to enterprise search embedded in team workflows, and Lovable is a step earlier. The signals appear in a fixed order: pricing shifts from seats and credits to platform tiers, compliance certifications land, forward-deployed hiring accelerates, and only then does the ARR mix move. The failure mode is equally observable: an enterprise tier launches and the mix does not move within four quarters. That company is a Velocity Racer carrying a System Builder price, the single most common mispricing in this market.

02 → 01. The model flip. A Velocity Racer that scaled fast holds something labs covet: proprietary usage data at volume. Building its own model converts distribution and data into a defensible asset while cutting cost and dependency. Cursor shipped its Composer models after scaling on coding usage. The honest nuance for a technical room: Composer was built on an open-weight base rather than trained from scratch. That’s precisely the point. Open weights made the arrow affordable; the proprietary data and distribution made it worth taking.

04 → 03. The embedding flip. An agent that delivers outcomes accumulates the customer’s workflow data, then turns into the system of record the customer runs on, trading per-outcome billing for embedded retention. Sierra or Notch are the cases to watch, expanding from resolutions into an Agent OS. Discipline requires saying it plainly: no Outcome Engine has yet completed the flip to platform-dominant revenue. It is an emerging pattern, not a proven one, yet!

The threat paths

01 → 02. Labs eating the app layer. A lab owns the model; capturing the end-user application is the highest-margin way to monetize it, so it moves down the stack against the apps built on top of it. Anthropic shipping Claude Code against Cursor and Lovable is the live case, and the damage is measurable: Cursor’s share of AI coding spend fell from 41% to roughly 26% in under a year on Ramp’s data, with Anthropic taking half the category. The distribution variant is more brutal still: Apple Intelligence, Gemini in Android, Copilot in Windows can erase a Velocity Racer niche at the OS level overnight. The defense is the golden path itself: embed into workflows before the lab arrives.

01 → 05. Labs reselling their own infra. A lab already runs one of the largest infrastructure footprints in AI, and can turn any layer of it into a product: OpenAI’s AgentKit against LangChain, LangSmith, and the eval tools; Mistral Compute against the GPU clouds. The squeeze lands on pure-play inference and orchestration, whose largest suppliers become their competitors. Demand concentration makes it existential: when a handful of labs are most of your revenue, any one of them integrating down can end you. This is why OpenRouter’s risk profile differs structurally from Together’s or Fireworks’: every new competitive model adds traffic to a router and subtracts pricing power from an inference provider.

Legacy incumbents → 03. The counter-migration. The one arrow that enters the map from outside it. Intercom with Fin, Salesforce with Agentforce: the incumbent bolts agents onto its installed base, and it does not need to win. It needs to be good enough to block the System Builder from entering or expanding inside the account. This is the existential clock on breed 03, and the reason the window to entrench is the narrowest constraint in the map.

The arrows that do not exist

Some migrations are structurally unlikely, and knowing why sharpens the map. A System Builder does not become a Velocity Racer: enterprise embedding does not reverse into self-serve simplicity, and no one has ever un-implemented themselves into virality. An Intelligent Hardware company does not migrate into software families: its moat, certifications, clearances, manufacturing, is precisely the asset that does not transfer. And almost nobody migrates into breed 01 from outside breed 02: without proprietary usage data at scale, the arrow has no fuel. When a pitch claims one of these arrows, the burden of proof is on the pitch.

What migration does to price

Three patterns recur. The multiple re-rates before the revenue mix does: investors pay the destination breed price on credibility, which is where both the upside and the mispricing live. Migration announcements cluster around fundraises: the enterprise-pivot narrative and the round arrive together, so discount the narrative and audit the mix. And the contested exits sit on unfinished arrows: Outcome Engines will be priced as software or services depending on where automation rates settle. The practical rule: classify the company by its current dominant revenue engine, then underwrite the arrow. The breed tells you what you are buying today. The arrow tells you what you are paying for.

The European read

European sovereignty in AI is real but uneven, and the map is sharpening rather than filling in. Europe is now genuinely strong in three breeds, and the white space in the other three is just as instructive.

Where Europe is strong.

European sovereignty in AI is real, but it’s uneven, and this year that unevenness is easier to see than it was twelve months ago.

Frontier Labs: Mistral, AMI Labs, Recursive Intelligence, Fundamental, Gradium, ElevenLabs, Black Forest Labs, Ineffable, Stability, Bioptimus, and a deepening bench behind them.

Velocity Racers: Lovable going from zero to $400M ARR in under two years is the archetype for the whole breed globally, with DeepL, Legora, Synthesia, and Granola right behind it.

System Builders is where the AI Europe 100 2026 actually centers: more than 40% of this year’s 100 companies. That’s not an artifact of how the list was put together. Europe’s enterprise market rewards exactly what this breed does well: regulated verticals, compliance-heavy procurement, and legacy software fragmented by country and language, which buys local entrants time to dig in before a global winner shows up.

The cohort is wide. Attio is rebuilding the CRM from London. Wordsmith raised a $70M Series B to become the system in-house legal teams run on, with BT, Canva, and the Financial Times among its customers. Maki runs enterprise hiring end-to-end on top of the ATS for H&M, BNP Paribas, and FIFA, while Nevis, founded by Revolut alumni, raised $35M from Sequoia to build the operating platform for US wealth advisors. Trawa is doing to energy procurement what these companies did to legal software, and Primo is doing the same for SME IT.

Intelligent Hardware is Europe’s most uneven category right now. European defense, security, and resilience startups raised a record $8.7B in 2025, up 55% year over year, with AI underpinning 44% of that funding (Dealroom / NATO Innovation Fund). The top of the breed has re-priced to match: Helsing is raising at an $18B valuation, Harmattan is at $1.4B with Dassault Aviation leading and French and UK MoD orders within two years of founding, and Quantum Systems is close behind.

The frontier is consolidating, not fragmenting. Cohere’s acquisition of Aleph Alpha, announced in April 2026 and still subject to regulatory approval, would create a $20B transatlantic sovereign-AI group with Cohere holding roughly 90% of the combined entity. If it closes, Mistral becomes Europe’s only independent frontier LLM developer left standing. Sovereign consolidation looks like the exit path for frontier labs that don’t win the race outright.

Then there’s the white space. Stack Enablers remain thin: Nscale is the standout, Hugging Face is French-founded but US-incorporated, Fluidstack pivoted to the US, and the tooling layer barely registers.

Outcome Engines barely exist at scale either. Parloa is nearly alone at the top, though a small cohort is forming underneath it.

There’s a bigger capital problem behind all of this. Europe matches the US on talent and on startup creation, and still trails it by a wide margin on late-stage capital, roughly $12B against $141B. The same gap runs through defense: even after a record year, US investors still account for the large majority of NATO-ally defense-tech ventures, much of it concentrated in exactly the late-stage rounds Europe can’t yet write on its own. Europe is good at sourcing deals and building companies. Getting those companies to scale globally is the piece still missing.

That gap is real, and so is the momentum behind it. Europe isn’t scanning for signs of life anymore, not in the categories that work. It’s picking among four families where it already has category leaders, while a hundred companies fight over what the next generation of those leaders looks like.

Outside the frame, deliberately. Three deliberate exclusions. AI biotech spends like a Frontier Lab but exits on pharma economics. Crypto × AI plays by different rules and deserves its own map. AI consulting is a massive market, mostly outside venture, yet it shapes how every breed here gets adopted.

Disclaimer: Headline and its affiliated funds may hold positions in companies mentioned in this post. This post is for informational purposes only and does not constitute investment advice. Figures cited reflect publicly available information as of July 3, 2026.

Today, we’re excited to announce our investment in Taxwire, which has raised $25 million. We co-led the Seed with XYZ Ventures and quickly doubled down as the company scaled 5x in under eight months from when we first met the team.

Sales tax sits in an awkward middle ground: too complex for SMB tools, too operational for enterprise software, and too critical to get wrong. Mid-market finance teams end up owning the outcome without the right expertise or tools, leading to overpayment, under-filing, audit risk, and endless manual work.

From personal experience, the dominant platforms charge enterprise prices for complexity and still don’t solve the problem. Co-founders Andrew Rea and Steven Schmatz saw the same gap and built Taxwire.

In the last year, Taxwire has served over 300 customers, recovered millions in overpayments, enabled cross-border compliance, and displaced more than 150 legacy systems. This round fuels the next phase of growth.

Our belief is that the hardest part of sales tax isn’t calculation or filing, it’s everything before and after. It’s structuring messy transaction data, mapping inconsistent product catalogs to tax rules, managing exemption certificates, and handling state notices. This is where legacy systems break and where Taxwire is focused.

Taxwire is built at the transaction layer, where tax actually fails. Their proprietary engine spans over 20,000 jurisdictions and more than 100,000 rules, delivering real-time, rooftop-level accuracy. Combined with deep integrations across ERPs, billing systems, and payment processors, plus end-to-end workflows from nexus determination through audit support, the result is a system that works in practice not just in theory. This allows Taxwire to work across industries, geographies and actually manage services, not just calculate. Taxwire works across multiple industries including commerce, software, marketplaces, manufacturing and more.

One customer mentioned that they provided “ tooling to track nexus exposure, manage registration paperwork, and keep the company proactively informed about when they were approaching nexus thresholds. The hands-on service and willingness to assume ownership were key decision factors. When new registrations need to be filed, Taxwire provides significant lead time before any nexus-triggered requirements, so the customer is never caught off guard. They integrate with customer’s Stripe instance, and tax amounts also feed into their reporting and ERP systems.” They also described the Taxwire team as “too impressive to ignore.” Another customer mentioned that prior to Taxwire they paid Stripe, Avalara and outsourced accountants to reconcile things that would take 2 full business days.

While incumbents like Avalara and Vertex grew through fragmented systems, and newer entrants like Numeral and Anrok remain more verticalized, Taxwire is building a unified platform across industries, geographies, and use cases from day one.

A key differentiator is embedded intelligence. Taxwire learns from past invoices, applies tax logic automatically, and creates deterministic outcomes with built-in audit trails. Manual workflows like exemption management or notice handling are automated directly within finance teams’ existing systems.

Just as importantly, Taxwire is not just software, it’s a fully managed compliance solution. The team handles nexus, registrations, filings, remittances, and notices end-to-end. Finance teams don’t adopt another tool; they hand off the problem entirely.

The timing is critical. Enforcement is increasing, global commerce is expanding, and tax rules are becoming more complex. Meanwhile, legacy tools are falling short, and new technology makes true automation possible. Every company operating across jurisdictions needs this to work and today, most don’t.

Taxwire is led by Andrew Rea and Steven Schmatz, combining speed, customer empathy, and technical depth with deep domain expertise from leaders across Deloitte, EY, TaxJar, Avalara, Zamp, and Abacum. That mix of high-slope builders and true tax expertise is rare,and essential.

Sales tax compliance isn’t optional, it’s core infrastructure. The kind that keeps companies out of audits and operations running smoothly. We believe Taxwire is building the platform that will power every dollar of sales tax collected, and we’re excited to partner with them on that journey.

Today we are thrilled to introduce two people who will help build and strengthen those relationships alongside Estelle Mamane, our Head of Investor Relations Europe: Iñigo Garcia Gordobil, joining as Partner and Global Head of Investor Relations, and Morgan Dickson, joining as Director of Capital Formation.

Iñigo Garcia Gordobil

We at Headline have known and worked with Iñigo for over 13 years and are thrilled to welcome him to the team. He joins us from Base10, a technology-focused venture capital firm, where he was a partner and the firm’s first Head of Investor Relations. Before Base10, he focused on venture and growth technology investments at QIC, one of Australia’s largest institutional investment managers.

His path to investor relations began at Cambridge Associates, where he spent nearly a decade evaluating venture capital and growth equity managers, including Headline. As a result, he developed an acute understanding of the firm initially as an allocator: through data, repeated conversations, and pattern recognition built across hundreds of managers globally. It was at Cambridge, while leading the allocator’s Latin America private investment research practice, that he first met Headline co-founder Mathias Schilling, who was building what would become Headline’s South American presence. They stayed in touch over the years that followed.

At Headline, Iñigo will oversee all LP relationships and looks forward to partnering with Headline’s leadership, along with existing and future LPs, in building the next era of the firm:

“From my early days as an LP, I’ve respected Headline. The team is exceptional and the track record speaks for itself. But in addition, very few firms operating today have managed to maintain a commitment to being truly global – and benefiting from the competitive advantages that presents – while executing with a real understanding of local market nuance. I see significant potential for Headline’s continued evolution and am excited to now be a part of the journey.”

Morgan Dickson

Morgan Dickson and Iñigo are no strangers to each other: the two attended the same small university in Lugano, Switzerland, and first connected through the close-knit alumni network. Both later moved to the Bay Area and have stayed in touch in the years since, professionally and personally.

Morgan comes to Headline from one of the world’s largest alternative asset managers, where she spent over three years managing relationships with single- and multi-family offices, endowments, and foundations in the Western US and Canada. She also spent four years at Hall Capital Partners, a prominent OCIO based in San Francisco. While at Hall, Morgan was a member of the Portfolio Management and Investment Operations groups, helping to support the firm’s ultra-high-net-worth family, endowment, and foundation clients.

Her responsibilities at Headline will include supporting Iñigo with all aspects of the LP lifecycle as well as other firm value-add initiatives.

“From my very first conversation with Mathias, I knew he had built a special place. I’m incredibly excited to be a part of the team. There’s a lot of momentum here at Headline, and I look forward to making a meaningful contribution to the firm’s next chapter.”

Looking Ahead

LP relationships always deserve dedicated attention not only during fundraising cycles. Iñigo, Morgan, and Estelle represent our commitment to delivering an IR function worthy of the partners who have believed in us, and the ones we hope to earn the trust of in the years ahead.

Welcome to both!

Compliance requirements have multiplied at the same time. More than 100 countries now mandate real-time e-invoicing and digital reporting. Companies are scaling across borders earlier and faster than ever before. Meanwhile, boards expect tax functions to do more with less - obligations are growing, headcount is not. Every other function is being transformed by AI, while indirect tax has struggled to implement production-ready AI workflows.

The result is a structural mismatch: high-volume, multi-jurisdiction businesses running on tax infrastructure designed for a slower, local era. The category is ripe for reinvention.

In recent years, Fonoa has proven itself to be a global leader in indirect tax for modern enterprises globally. Its modular platform is built on a single data model, designed for the high-volume, real-time needs of some of the world’s largest enterprises including Uber, Netflix, Spotify, Booking.com, Zoom, Dell, Canva, Intuit, Lenovo, among others. The global indirect tax software opportunity alone is approximately $30 billion today. The services market opportunity is several times larger. Both budgets are directly addressable for Fonoa.

Headline is proud to lead Fonoa’s $110 million Series C, joined by new investors Eurazeo and Forestay and alongside existing investors Index Ventures, OMERS, Coatue, and Dawn Capital.

Why Fonoa

Four reasons drove our conviction:

Founders who lived the problem at scale and are driving the industry forward globally. Davor Tremac and Filip Sturman saw the indirect tax problem firsthand at Uber, where Davor was General Manager for Southeast Europe and Filip a Senior Manager. They watched legacy software fail to support Uber’s high volumes of and real-time transactions across 70+ countries. Fonoa is the platform they wished they had. Rob van der Woude - Uber’s former Head of International Tax and a former Uber board member - joined as Chief Tax Officer in Fonoa’s earliest days.

Founded in Zagreb, Fonoa is now a global company with regional hubs in the US, EU, UK, and Australia. It has earned the trust of the world’s largest enterprises and attracted world-class talent across every team. Davor and Filip are true pioneers driving the industry forward on a global scale.

A tax operating system built on modular infrastructure with true data intelligence. Unlike the old way of patching together many point solutions with siloed data, Fonoa’s modules cover tax determination, tax ID validation, e-invoicing, invoicing, returns, and data sharing on a single platform with one audit trail and data model. It supports tax determination across more than 190 jurisdictions, validates tax IDs in over 100 countries, and processes more than a billion transactions annually. The result? A dramatic reduction in manual burden alongside meaningful cost savings and ROI for customers.

A blue-chip customer base, deep customer love, and strong net revenue retention. Fonoa serves leading enterprises across technology, financial services, electronics, manufacturing, retail, transportation, and beyond in every major market. These are companies operating at transaction volumes that overwhelm legacy tax solutions. Many came to Fonoa after multi-year, multi-million-dollar attempts to make incumbent software work or to build a solution internally. Customers report up to 90% faster tax calculation, meaningful accuracy gains, substantial reductions in compliance headcount, lower outsourced services costs, and meaningful vendor consolidation across the stack and various jurisdictions.

A platform purpose-built for autonomous tax. Fonoa is building the operating system for autonomous tax: AI that tracks every rule, acts on every obligation, and proves every decision. Agents monitor obligations, populate returns, catch anomalies, and assemble audit packs in seconds - running on clean, structured, versioned tax data from day one, with no legacy codebase constraining what can be built on top. The system does the work while humans assess and make the calls.

Looking Forward

Alongside the Series C, Fonoa is acquiring Indirect Tax Edge, PwC’s enterprise tax compliance platform used by hundreds of multinational customers to manage VAT and GST reporting, e-filing, transactional data management, and tax analytics. PwC is a world leader in indirect tax services, and this acquisition is a strong endorsement of what the Fonoa team has built.

The Series C and the PwC Edge acquisition mark the start of Fonoa’s next chapter. Fonoa is defining AI indirect tax for global enterprises.

The challenges facing Europe are well-documented. However, as investors with a foot in both the US and Europe, we believe that we are well-placed to comment on the progress that Europe is making. Europe faces a delicate balancing act, which involves moving closer to some parts of American capitalism while maintaining its distinctive character. We see early signs that this can happen—in fact, it is already happening. As investors, we know that we can only play a small part in this process, but the current European investment landscape leaves us excited for the future.

1: The Problem

Over the past 30 years the US has moved ahead of the EU on the most comprehensive measure of living standards. “Actual individual consumption per capita” includes the income that people get from their job, but also includes the benefits that people get from public services like free education and healthcare. Not long ago Europe was ahead of the US. Now it is behind.

The same pattern holds across other economic variables. In the year 2000, the average worker in the (current) European Union produced $54 of economic value for every hour that they worked, measured in today’s prices. The average American worker produced $59 – a slight advantage. But today, Europeans produce $67 of per-hour value while Americans produce $84.

The gap is even clearer in markets. The US dominates global equities: American companies account for roughly 63 percent of the global MSCI ACWI index. US companies also command higher valuations. The price-earnings ratio of the MSCI USA is 35 percent higher than the MSCI Europe. Investors in the US markets expect stronger earnings growth in the future.

Closer to venture capital, the gap in fast-growing companies is larger still. In the past 50 years no company with a market capitalization above €100 billion has been founded in the EU. European venture capital has grown significantly over the past decade, but it remains structurally smaller than the US market. According to a recent paper by the Bank of France, America’s venture capital market is about 10 times the EU’s.

2: Diagnosing Root Causes

A growing number of Europeans are taking these problems seriously. In recent years researchers have published a range of reports diagnosing the problem. This includes Mario Draghi’s report in 2024, as well as Enrico Letta’s single-market report, the European Commission’s Competitiveness Compass and the new Savings and Investments Union. The conclusions are all broadly similar. Europe has too much fragmentation, too little risk capital and too much regulatory complexity.

The most obvious issue is capital. Europe is not poor. It has large household savings pools. In 2024 EU households saved well over €1 trillion (see chart below). This pool of savings could be put to use in turning fast-growing European companies into world-beating ones. Indeed, household savings are the ultimate source of risk capital.

However, too many of those European savings sit in bank deposits, rather than being channeled into productive investment. This is partly a regulatory question: in parts of Europe pension funds are encouraged to put their assets into safe but low-returning assets, such as government bonds. It is also partly a cultural question. Americans are far more likely to be retail investors than Europeans, who seem to be more risk-averse on average. Without changes, European companies may continue to face a shortage of scale-up capital.

Europe is a continent but it is not a market. In fact, it is made up of many different markets operating in parallel. The promise of the single market was to abolish trade barriers between countries, allowing businesses in one country to provide goods and services in another. This promise remains fundamentally unfulfilled. Yes, there is free trade in goods – for example, wine produced in France can easily be sold in Italy. Yet Europe is a long way from free trade in services, from banking to the law. A French lawyer, for instance, cannot easily practice in Italy, even if fluent in the language. Europe was making real progress towards this goal during the 1990s and 2000s, but has made essentially zero gains in the past decade.

Policymakers across the political spectrum broadly agree on the diagnosis. The harder and more consequential question is whether that consensus can translate into coordinated action.

3: The Case for Reform

The skeptical case for European reform is easy to make. France is the obvious example. Emmanuel Macron’s pension reform raised the retirement age from 62 to 64, but it triggered intense opposition and was later suspended.

But the opposite cases matter more than they get credit for. Reform is politically possible in Europe, and when it happens, it changes economic outcomes.

Greece is the strongest example. Fifteen years ago, Greece was the symbol of European fragility. Today it is one of the more convincing reform stories in the euro area. In recent years the country has made enormous strides in digitizing public services, helping to deliver them more transparently and at lower cost. It has also enacted large-scale labor reforms, making it easier to hire and fire. Product-market reforms have increased competition in a range of different industries.

The results are clear. Twice in the past four years, The Economist has named Greece “economy of the year”. Greece has delivered cumulative growth of about 20 percent since 2021, while public debt has fallen by more than 55 percentage points of GDP from its peak (see chart below).

The Baltics are another case study. Estonia, Latvia and Lithuania went through brutal adjustments after the financial crisis. These countries did this partly in order to meet the various macroeconomic targets that were necessary for them to join the euro, which all of them have now done. It was painful, but it preserved competitiveness and created a platform for later growth. Estonia in particular has since built one of Europe’s most successful startup ecosystems relative to population, with very high unicorn and VC density per person.

We focus on Greece and the Baltics because it helps overturn a common narrative about the EU: that it is in some sense “unreformable”. This is not true. They are also not the only examples. These cases remain underappreciated in mainstream economic commentary.

4: Early Signs

Europe’s turn may already be underway. From our interactions with European investors, founders and policymakers, we sincerely believe that people know that something needs to be done. The sense of urgency is palpable. We cannot quantify this, of course, though it is obviously important.

Alongside Greece, countries like Cyprus and Portugal are demonstrating that reforms are possible. Indeed for the first time, those three countries have a lower unemployment rate than the Nordics. This is a truly remarkable achievement that would have seemed impossible just a few years ago. In fact, the European economy’s overall labor market has clearly improved relative to the US. A decade ago the EU’s unemployment rate was 4.4 percentage points higher than the US’s. Today it is just 1.7 points higher.

Europe’s investment ecosystem is attractive, in our view. In the past year European public companies have offered investors a total return of well over 20 percent. Recent coverage has described “reasons to be cheerful about European tech”. For the first time, there is net movement of technology workers away from the US and towards the EU. Venture capital spending is rising fast. Paris and London, in particular, are global centers of AI.

Our view, though, is that expectations for European companies are still constrained by the belief that the European market has a low ceiling. If reforms proceed as we expect, that ceiling will rise higher and higher.

Final Thoughts

The stakes are high. Europe’s challenge is about maintaining the European project. The continent needs to finance defense, aging, welfare states and strategic autonomy without drifting into fiscal stress.

The pessimistic view is that Europe is too fragmented, too slow and too politically constrained. The optimistic view, to which we subscribe, is increasingly credible.

For investors, this creates a clear framework. Europe is still under-scaled, under-capitalized and under-valued relative to the US, but also has large savings pools, world-class talent, strong research, and policies which are improving. If even part of the reform agenda is executed, Europe could be in for a great decade.

Congratulations to Founder and CEO Chad Janis and the entire Grüns team on this exciting next chapter. With Unilever behind them, a daily ritual that millions already love is about to reach many more.

Why Grüns

Headline has been investing in consumer brands for more than 30 years, backing companies like Sonos, The RealReal, Farfetch, and Groupon from scrappy startups to household names woven into everyday life. That history produces a specific kind of pattern recognition for what generational consumer brands look like early, often before the broader market has caught on. It shaped how we evaluated Grüns, and what we found was remarkable.

The brand first came on our radar when our early-stage investment team stocked our office with Grüns’s signature, colorful packs. We evaluated the company for our new, nearly $1 billion Growth Fund last year and the signals were clear: in just a year post-launch and with an exceptionally lean team, Grüns had scaled to nearly $300 million in annual run-rate revenue with a constantly improving product and a devoted customer following. Remarkably, that scale came with discipline: customer acquisition was efficient, gross margins were best-in-class, and the business was profitable.

The deeper we looked, the more our conviction sharpened:

The product is a daily habit, not a one-off purchase. People take vitamins every day – and they eat gummies because they taste great. Grüns combines wellness, taste, fun, and convenience into a daily ritual. 96% of customers enjoy their gummies four to six times a week and 80% of customers take them daily. This is the kind of behavior that produces exceptional loyalty: Grüns cohorts retained at rates that would be enviable to many software companies. That level of integration into everyday life is rare and difficult to replicate.

The business model is durable. Unlike single-ingredient supplements, Grüns is built around staple consumer needs: nutrient support, immune system support, cognitive support, and energy support. The company sells through DTC, Amazon, and national retailers, giving it diversified distribution and multiple vectors for growth.

The brand is distinct and self-sustaining. The umlaut is instantly recognizable. The teddy bear is iconic. And critically, the brand does not depend on a singular influencer or trend. Instead, Chad and his team are marketing wizards who have built a voice that speaks transparently and a brand that people fall in love with.

The form factor is a platform. A pack of teddy bear gummies may seem simple, but there is real value in this unassuming packaging. This format drives loyalty, increases repeatability, and extends naturally into new benefit areas and demographics. When we first invested, Grüns had one product: the adult green gummies. Since then, they’ve launched Kids, Sugar-Free, Nütrops (cognitive support), Immün (immune system support), Jüced (energy support / pre-workout), and a number of limited-time flavors (Grinch Punch was a Headline office favorite). Each new product opens up a new consumer segment, increases average order value, and gives existing customers more reasons to stay.

They’re a best-in-class team. Much of this comes back to Chad. He is direct, relentless, and deeply focused on product quality, execution, and bringing out the best in people. He has assembled some of the finest talent in this industry and they, in turn, have rallied around a shared commitment to product quality and customer impact.

What Happened Next

We led Grüns’s Series B in May 2025. From the start, the Grüns team exceeded our expectations. And we went in with high expectations.

Within weeks of launching with their first retailer, Grüns became the number-one supplement brand on its shelves. The retailer was so impressed that they accelerated their timelines to stock additional SKUs ahead of their standard annual reset. Other major retailers followed, expanding Grüns into thousands of storefronts across the country, including Target, Walmart, Sam’s Club, and Costco.

Looking Forward

Grüns is the first exit from Headline’s IV Growth Fund, which launched with a new dedicated team in late 2024. The strategy is built around a concentrated fund, a small and focused team, and the full weight of Headline’s global platform and relationships. The Grüns outcome is an early and meaningful proof point for that approach, and for the kind of conviction-driven consumer investing that has defined Headline for more than three decades.

We asked Chad to reflect on the partnership:

“Headline has been the ideal partner in our scale-up journey. You can tell a lot about an investor by how they show up when things are moving fast. Headline matched our intensity from day one. They believed in the product, they appreciated the team, and they understood our purpose. Most importantly though, they’re visionaries. You need visionary partners who exercise unrelenting belief and support. I’m grateful Headline was a part of our story.”

Congratulations again to Chad and the entire Grüns team. The world is about to eat a lot more gummy bears.

Most people talk about Consumer AI in terms of features. At Headline, we think about it in terms of human needs.

Zooming out, you start to see a clear hierarchy:

• AI that helps us live

• AI that helps us stay safe

• AI that helps us belong

• AI that helps us grow

• AI that helps us become who we want to be

In other words — Maslow’s hierarchy is quietly becoming the roadmap for Consumer AI.

Maslow’s hierarchy explains human motivation, from survival to self-actualization. Consumer AI is following the same arc: from tools that meet basic functional needs to systems that augment creativity, connection, and identity. The next generation of AI startups will climb this pyramid, one layer of human need at a time.

The startups that win this decade won’t just build tools, they’ll build companions, copilots, and catalysts for human potential.

Physiological Needs – AI for Daily Survival & Efficiency

• We’ve moved from calories to computation – AI meets the needs of modern survival: convenience and control

• Household complexity has exploded (subscriptions, devices, diets, schedules), while models are now good enough to orchestrate daily life, turning fragmented services into “autopilot for living”

• AI shifts consumers from reactive decision-making to proactive optimization – reducing cognitive load across food, money, and home systems, and unlocking meaningful hours of regained time

• We’re excited about AI meal planning, home energy optimization, budget copilots, AI-powered grocery / recipe automation

Safety Needs – AI for Security, Health, and Stability

• AI becomes the new insurance layer – always-on guardians for our health, memory, and money

• Scam sophistication, data exposure, and digital complexity have outpaced human ability to self-protect, but multimodal AI is finally capable of detecting and protecting from new threats across channels, devices, and finances

• AI delivers ambient safety – continuous, personalized monitoring that catches threats before consumers even notice them, making security preventative rather than reactive

• We see opportunities in AI cybersecurity for consumers, personal data firewalls, health monitoring, legal / financial copilots

Belonging & Love – AI for Connection & Identity

• The next social network won’t connect people – it’ll connect their digital selves

• Rising loneliness and shrinking IRL social graph density meet emotionally capable LLMs capable of memory, personality modeling, and emotionally congruent responses, along with models that can form and maintain relationships

• AI is a new social substrate, helping people find affinity groups, express identity, and maintain companionship

• We’re excited about AI companions, community formation, interest-based matching, emotionally intelligent agents

Esteem – AI for Mastery, Self-Image, and Productivity

• AI is becoming the ultimate self-esteem machine – amplifying output and recognition

• We’re at the moment where AI can evaluate, coach, refine, and create at near professional levels, democratizing expertise

• AI turns amateurs into semi-pros, giving users both the ability and confidence for self expression, fast learning, and max productivity

• We’re intrigued by AI creativity tools, personalized learning copilots, social capital engines (AI content co-creators)

Self Actualization – AI for Purpose & Fulfillment

• The highest layer of the AI pyramid is not automation, it’s actualization. AI as mirror, muse, and mentor

• Consumers now seek not just efficiency, but also meaning. AI has become capable of long-horizon reasoning, personalized guidance, and reflective dialogue – the core ingredients of coaching and personal transformation

• AI unlocks scalable self-development that previously required expensive and scarce human experts

• We see opportunities in personalized education, life planning, AI-guided wellness or therapy, creative exploration tool

Vinay Reddy (Stanford MBA ‘27) was also a collaborator on this report.

Procurement is one of the most important functions within an organization, however, it has been overlooked from an operating system and workflow perspective for companies of all size and scale. Teams have long operated with thousands of supplier interactions stitched together across email, ERPs, and spreadsheets, with forced complexity using tools that were never built for the pace of modern trade. We have long believed that the next generation of direct goods procurement will win by meeting customers where they are and enabling them to have a frictionless experience in their most high frequency workflows by removing the manual coordination layer, and giving operators automation, visibility, and control.

In the search for a team, product, and vision that had the shared philosophy, we are excited to announce our investment in Didero’s $30M Series A co-led with Chemistry and with participation from M12 (Microsoft’s venture fund) and continued investment from First Round Capital and Construct Capital. Didero is a New York based company deploying AI agents that operate inside the systems procurement teams already use, handling some of the core workflows such as supplier outreach and communication, order tracking, invoicing, PO management, and exception management, so manufacturers and distributors can spend less time chasing status, reconciliation, and more time driving strategy. When we met the founding team (Tim Spencer, Lorenz Pallhuber, Tom Petit) it was obvious that their experiences within global trade and supply chain deeply aligned with the DNA of Headline as a global firm with teams around the world.

Didero is showing what an “AI native” platform looks like in the real world: fast and true ROI, minimal implementation burden, and measurable operational impact. A few things truly stood out in the business to us. Firstly, manufacturers and distributors were adopting it across various industries with no hesitation. This ranged from pharma companies, to pet food manufacturers to construction distributors. This is something we definitely didn’t see possible prior to AI. The best procurement businesses had been built on discrete verticals rather than more broadly. The problems, solutions and onboarding we heard from customers was incredibly similar and required little to no modification. Secondly, companies quickly saw the ability to decrease headcount in cost centers in business around purchasing, supplier management and invoicing to focus their time and energy on scaling revenue. Lastly, their team shared our views that the core of what Didero has built is enabling the potential world of autonomous procurement, giving the potential for companies to better understand supplier concentration and distribution, constantly assess and augment supplier bases and streamline purchasing workflows to turn cash management cycles more quickly. We believe this will allow for procurement and supply chains to operate at their best.

The Didero team will continue to accelerate product and go-to-market, expanding beyond core procurement and into adjacent workflows over time. We are excited to partner with them in this journey in building a world class team and product to transform procurement and global trade. If this mission resonates with you, the company is hiring across enterprise sales, customer success, and technical roles as they continue to serve some of the best manufacturers and distributors around the world.

We’re continuing to see a real wave of ambitious founders building in NYC, where teams are iterating faster than ever, just a stone’s throw away from their customers, and a talent density that is contributing to this great city and beyond. If you are building something great, we would love to hear from you! You can reach us at taylor@headline.com and jerry@headline.com

Today, Fundamental is emerging from stealth with $255M in total funding, a $30M Seed and a $225M Series A led by Oak HC/FT, and publicly launching its flagship Large Tabular Model (LTM), NEXUS. The new capital will be used to scale compute, expand enterprise deployments, and grow the team across research, engineering, and go-to-market.

From the outset, Fundamental stood out as a rare combination of deep technical ambition and clear market relevance. The team is tackling one of the most persistent blind spots in modern AI: how to make sense of enterprise-grade tabular data at scale, reliably, under real-world constraints, and in the workflows where correctness matters. This isn’t incremental innovation. It’s foundational.

Since the Seed round, Fundamental has executed against the vision we backed early on, strengthening its technological edge, sharpening its focus on high-impact use cases, and positioning itself as a reference player in what we believe will become a core layer of enterprise AI.

The Missing Rail in Enterprise AI

In our recent thesis, “2026 AI Predictions”, we frame AI progress through the concept of rails. Foundational models trained on specific asset types form the underlying infrastructure for applications. When a rail matures, application adoption compounds quickly—often after an early period where results can feel underwhelming.

We’ve seen this pattern repeatedly: text reached maturity first with large language models; image and video followed; voice is now approaching a similar inflection point across consumer and enterprise use cases.

One rail, however, remains structurally underdeveloped: tabular data.

Despite sitting at the core of enterprise decision-making, tabular data has not benefited from comparable breakthroughs at the model level. Most deep learning progress has centered on unstructured, sequential data (text, images, video), creating a structural mismatch for relational, schema-heavy tables. In practice, enterprises are left with legacy methods or brittle workarounds (rules, feature pipelines, or lightly adapted LLM workflows) that struggle to scale reliably as data grows in size, dimensionality, and complexity.

Yet tabular data is where enterprise truth lives: structured, interconnected, sensitive, and often regulated. It underpins the highest-stakes decisions, where robustness, auditability, and precision matter more than impressive demos.

We believe the maturation of this missing rail will drive the next phase of enterprise AI.

Why We Invested at Seed: A New Model Class for a Clear Pain Point

Our conviction at Seed was straightforward: unlocking tabular data at scale requires a new class of models. This isn’t a product gap or a tooling limitation. It’s an architectural constraint.

Tabular data is relational, schema-dependent, and highly heterogeneous. Enterprise datasets combine numerical signals, categorical variables, timestamps, sparse features, and linked tables at scale. That structure is exactly where general-purpose models and classical methods tend to break down, especially as data becomes deeply relational, and the cost of error is high.

We believed incremental improvements, whether through rules engines, decision trees, or lightly adapted LLMs, would not be sufficient. What was required was a dedicated foundation: Large Tabular Models (LTMs), designed from first principles to reason over complex relational data while preserving reliability, explainability, and control.

Timing also mattered. By 2024-2025, enterprises had mature data infrastructure, clearer expectations around AI in production, and growing regulatory pressure around explainability and governance. At the same time, decision-making workflows were increasingly expected to become AI-driven. Yet the underlying data powering those decisions remained largely untouched by modern models.

Our Seed thesis was that LTMs would unlock disproportionate value in a relatively small number of decision-critical domains, and, over time, become a core layer of enterprise AI infrastructure.

The sectors are familiar, but the pattern is consistent:

• In financial services, tabular data drives risk assessment, fraud detection, pricing, and capital allocation decisions.

• In insurance and healthcare, it underpins claims adjudication, coverage decisions, and outcome prediction under strict regulatory constraints.

• In supply chain, logistics, and industrial operations, it governs forecasting, inventory management, and resource allocation.

• In SaaS, telecom, and marketplaces, it sits behind churn prevention, pricing strategy, and usage-based decisioning.

The decisions are high-stakes, and the tolerance for error is low. Existing approaches fail to generalize reliably.

Our conviction was that once a model could reason natively over complex tabular systems, these workflows would shift from brittle heuristics to scalable, AI-driven infrastructure.

Why Fundamental Specifically

1. A Highly Complementary Founding Team

LTMs are not a pure research problem. They require the ability to design a new model class from first principles, while simultaneously meeting the constraints of enterprise-grade deployment. Very few teams can credibly cover both dimensions. That’s where Fundamental stood out.

From inception, Fundamental brought together an exceptional technical core, including Wojtek Czarnecki, Gaël Varoquaux, Ben Nachman, and Alex Perez, with deep, complementary expertise across foundation models, tabular learning, causal inference, and large-scale AI systems. This concentration of technical depth matters because the category requires setting a new architectural standard.

Just as importantly, the team paired this depth with strong commercial and operational leaders. Jeremy Fraenkel (CEO) and Gabriel Suissa (COO) have shown the ability to translate frontier research into enterprise reality: articulating a new category, educating a nascent market, and engaging strategic customers operating under strict data confidentiality, governance, and procurement constraints.

The result is a rare level of complementarity: frontier AI research aligned with enterprise expectations from day one.

2. Global by Design, from Day One

Foundational infrastructure categories don’t stay regional for long. Standards get set globally, and credibility compounds.

Fundamental was global by design from day one: cross-continental footprint, international backgrounds, networks across the US and Europe, and early engagement with global financial institutions. That posture wasn’t aspirational, but it was strategically consistent with what it means to build a foundational rail.

3. The Ability to Raise Capital at Scale, with the Right Signal

We also knew this category would be capital-intensive. Building a foundational model requires multiple, often closely spaced funding rounds, and the ability to raise both in volume and with the right partners.

Fundamental demonstrated a thoughtful approach to capital strategy early on: not just raising capital, but raising it with the right partners. In infrastructure categories, the cap table is part of the product. It signals legitimacy, attracts talent, anchors future rounds, and can materially influence trust with enterprise customers during long sales cycles.

Across these three dimensions, team, global ambition, and capital strategy, Fundamental had the ingredients to define the LTM category.

Why We Doubled Down at Series A

Our decision to reinvest at Series A was driven by clarity, not momentum.

By this stage, the core risks we underwrote at Seed were meaningfully reduced. The technical direction proved credible. Market feedback confirmed the need for a dedicated tabular foundation. And the team executed at the pace and with the focus required for a long-term, infrastructure-level build.

With NEXUS, Fundamental is introducing a purpose-built tabular foundation model trained on billions of tables. Enterprises can integrate it into existing stacks with minimal effort, and the system is designed to learn underlying structure and dependencies without extensive manual feature engineering. It’s an important part of what makes category leadership plausible.

Over the past months, Fundamental has secured seven-figure contracts with Fortune 100 enterprises, applying the model to predictive use cases including demand forecasting, price prediction, and customer churn. Exactly the kind of high-stakes, high-ROI decisioning workflows we expected LTMs to unlock.

Series A is where the work shifts from feasibility to standard-setting. From building something that works to building something others build on. We believe Fundamental has a real opportunity to define this layer of the stack.

Looking Ahead: Making Tabular Intelligence a First-Class Citizen in AI

As AI moves deeper into decision-critical workflows, enterprises will require models that can reason reliably over their core data. And in most enterprises, that core data is tabular. Solving this challenge at the model level is a prerequisite for scaling AI beyond experimentation.

Fundamental is building toward that standard. By focusing on first-principles architecture, enterprise-grade deployment, and trust, the company is laying the groundwork for what we expect to become a reference layer for tabular intelligence.

We are proud to continue supporting the team as they build the infrastructure that will power the next generation of enterprise AI.

In an ecosystem driven by hype, separating real adoption from “AI tourism” is difficult. To cut through the noise, we partnered with the Early Builders community and surveyed 100 fast-growing European startups, from Pre-seed to Series B+, to understand which AI tools are actually embedded in day-to-day operations.

Here is what the data shows.

The Giants Are Compounding (and Co-Existing)

The “winner-takes-all” narrative doesn’t really hold at the application layer. In the AI era, the winners take most, but they often take it together.

ChatGPT remains the default cognitive layer, with 89% penetration. Claude has rapidly caught up, reaching 66%. Importantly, startups are not choosing one. They are paying for both.

Teams typically lean on Claude Sonnet for advanced reasoning and coding, while keeping ChatGPT for general-purpose tasks. This “multi-model by default” behavior drives unusually low churn for the leading AI platforms and allows revenues to compound rather than rotate. OpenAI and Anthropic doubled their ARR in H2 2025, reaching $20B and $9B, respectively, after tripling and quadrupling, respectively, in H1 2025.

The same pattern appears in developer tooling. Cursor leads with 76% adoption, yet GitHub Copilot still appears in 44% of stacks. Builders actively A/B test tools in production and accept duplicated seat costs to ensure best-in-class performance and reliability across edge cases.

Automation follows a similar logic. While multiple tools co-exist, European startups show a clear preference for technical depth: n8n has overtaken Zapier in our panel (56% vs. 31%).

“We try to select the best AI tool for each team’s use case. Using ChatGPT and company-wide custom GPTs for general-purpose tasks. Claude Code and agentic frameworks for engineering and Gemini Image generation for our content production.”

From Chat Interfaces to Agentic Workflows

While incumbents are compounding, the defining shift in 2026 is how AI is being deployed. The ecosystem is moving beyond chat interfaces toward agentic, multi-modal workflows. Tools that actually run processes, and not just answer prompts.

Coding & Data: Building Agent-Ready Infrastructure

Cursor and Lovable (76% and 35% adoption) have largely consolidated the interface layer. But consolidated doesn’t mean static. Recent Claude releases are gaining visible traction among developers, and Mistral has just announced a next-generation coding assistant, reopening competition at the model and system level.

As interfaces stabilize and models keep evolving, the real inflection point is shifting deeper in the stack. Startups are moving away from generic API calls toward infrastructure that makes LLMs reliable in production. Several functional layers are emerging:

• Structure: Companies are adopting .txt to push LLMs out of conversational ambiguity and into strict, machine-readable formats (JSON, Regex).

• Grounding: Hallucinations remain a key blocker. Linkup is emerging as a bridge between static models and the live internet, offering a search API that grounds agent outputs in verifiable data.

• Context: One-size-fits-all models are losing relevance. Dev teams are turning to Adaptive ML to fine-tune open-source models around enterprise KPIs, and to Lightbase to give agents contextual awareness of the entire codebase architecture.

Productivity: The Unbundling of the Workspace

For now, users are rejecting all-in-one assistants in favor of tools that do one job exceptionally well.

Granola reached 37% penetration by focusing on being a high-quality notepad for humans. This “niche-first” strategy is spreading across the productivity stack. Fyxer and Superhuman are taking over the inbox through automated triage and draft replies, while calendars are increasingly managed by dedicated tools such as Reclaim, and meetings are prepared by Himala.

History suggests this fragmentation won’t last. As tools like Superhuman expand from a wedge (email) into a broader team workspace, the category’s eventual winners are likely to consolidate use cases once they’ve earned consistent daily usage on their first wedges.

Customer Support: The Reliability Engine

Customer support is one of the few functions where LLMs already deliver clear value with very strong ROI. Adoption is accelerating even among small teams.

PolyAI (voice) and Pylon (mostly B2B startups) have become standard choices for handling complex interactions. The next phase is full autonomy, and Notch is at the forefront, resolving up to 70% of customer inquiries end-to-end (including complex edge cases) through omni-channel agents, without human intervention.

Marketing: Video Infrastructure and the Rise of GEO

Marketing is shifting on two fronts:

1. Video Infrastructure: As frontier models like FLUX from Black Forest Labs hit new performance levels, more value is being created at the application layer. Tools built on this infrastructure, such as the ad generator Higgsfield, are on a path to commoditizing high-end production costs.

2. Generative Engine Optimization (GEO): As users turn to AI answer engines instead of search, SEO is being redefined. Tools like Peec AI and Searchable help brands measure and optimize their “share of chat”, enabling marketers to understand how LLMs perceive and surface their brand. Ads inside ChatGPT will become a major acquisition channel, making GEA a core capability for GEO players and likely sparking a new wave of dedicated entrants.

Sales: From Systems of Record to Systems of Action

In sales, pre-AI tools like Lemlist and FullEnrich remain foundational for sequencing and data enrichment. The real shift, however, is happening in execution, where teams are experimenting with different approaches to automate the sales job itself.

Attio is an excellent example, implementing AI action features into its system of record and turning it into a system of action. In parallel, there are use-case-focused plays like Karumi, which replaces static “Book a Demo” flows with interactive video agents available 24/7, or Enginy, which deploys autonomous outbound agents that manage the full prospecting loop.

Finance: From Spreadsheets to AI Teammates

In finance, static spreadsheets are becoming obsolete. A new category is emerging around the “finance operating system”, with different wedges for different customer segments.

Payflows is rebuilding the financial stack for larger scale-ups around autonomous agents that manage discrete workflows. Rillet has similar ambitions for mid-market companies, while Hyperline addresses revenue management and billing complexity, freeing finance leaders to spend less time reconciling and more time making strategic decisions.

The Road Ahead

Adoption velocity is unprecedented — and reversible. The recent spike in Gemini 3 usage illustrates how quickly buyers are willing to switch when performance improves.

For founders, the challenge is no longer adding AI features. It is building organizational agility. The next generation of winners won’t be defined solely by model quality, but by their ability to integrate AI deeply into daily workflows, while staying flexible as more capable agents emerge.

The respondent panel comprised 100 European startups across eight countries. Each company reported the tools it uses across the following categories: LLM co-pilots; knowledge management and workflow automation; personal productivity; coding and application development; marketing; sales and GTM; customer support; finance; and HR.

For each tool cited, respondents also estimated internal penetration, defined as the share of employees using the tool relative to total headcount. The penetration rates shown above reflect the proportion of companies that reported using each tool.

Rails mature, apps explode, reality hits, then adoption compounds.

My core view is simple:

When rails mature, applications emerge fast. In the short run, outcomes disappoint, but operational maturity then quietly compounds adoption.

Rails are the foundational models trained on specific asset types: text, video, voice, tabular data, and, more recently, the physical world.

Every AI wave follows the same adoption pattern:

Rails mature → Applications emerge → Magic demos → Deceptive outcomes → Operational maturity → Scale → Authoritative capabilities.

Where we are today:

• Rails being built: tabular data, scientific research (bio, materials), the physical world

• Magic effect: video and voice-based use cases

• At scale: the most mature text-based applications like coding, customer support, scribes, and drafting

2026 is the year this cycle becomes visible across multiple rails, enabling agents to switch from assistive to authoritative capabilities.

Authoritative is when agents can autonomously execute end-to-end tasks inside real systems of record with permissions, logging, and measurable outcomes.

This fundamentally shifts the productivity paradigm: AI becomes a decisive actor, able to arbitrate between different paths, executing actions in the back-office, continuously running and optimizing parts of the organization 24/7.

1. Text rails are mature: Enterprise workflows are the battleground

In most enterprise deployments I see, text foundation models are no longer the bottleneck. They are strong, fast, and cheap enough for production. The hard part is everything around them: integrations, permissions, and what happens on edge cases.

That’s why competitive advantage has shifted from “better prompts” to a range of interdependent capabilities:

• Deep integration with systems of record

• Measurable ROI

• Reliable handling of edge cases

• Compliance and security readiness

• End-to-end execution

• High switching costs

Coding and customer support are the two enterprise use cases where most of these requirements are already being met, and usage, as well as spend, are scaling fast.

Why coding will remain the #1 killer use case

Coding has the cleanest feedback loops in enterprise software. You can measure time-to-ship, defect rates, cycle time, incidents, and review throughput. Early copilots delivered the magic effect: autocomplete felt like a superpower. Then reality hit: hallucinated APIs, inconsistent style, security concerns, and “looks right, breaks later.” The first verdict was clear: powerful, but not trustworthy.

Maturation is now obvious. In 2026, winning coding products will:

• Understand full codebases, not isolated files

• Run tests, diagnose, and fix failures

• Enforce security, licensing, and dependency policies

• Operate inside enterprise toolchains (CI/CD, secrets, observability)

• Produce auditable changes and reviews that teams can trust

Coding shifts from suggesting code to shipping software safely and consistently.

Why customer support will be the #2 enterprise AI winner

Support is also highly measurable: resolution rate, time to resolution, escalation rate, CSAT, and cost per ticket. The magic effect was chatbot-based deflection.

Once again, the disappointment followed: failures on edge cases, stale or incomplete knowledge, poor escalation logic.

What maturation looks like in 2026 in my view:

• Agents assist with grounded answers and citations

• Agents that execute end-to-end actions (refunds, replacements, plan changes)

• Quality monitoring and auditable outcomes

• Strict guardrails for permissions, logging, and escalation

This does not happen instantly, though. Enterprise AI customer support is not plug-and-play.
It requires deep integrations (CRM, billing, identity, order management) and operational design (who approves what, when, and why), which is generally enabled by customization via services teams or forward-deployed engineers.

What’s exciting is that this is starting to work in production. We’re still in the early days, and it feels difficult to imagine any competitive enterprise company not adopting AI for CX/CS in a few years. This is the beginning of a budget expansion cycle driven by proven ROI.

2. 2026 is the year video rails mature: Video applications explode

Video is where text was a few years ago. The magic effect is already here: “I can generate a scene from a sentence.”

And the disappointment is here too: character inconsistency, broken brand coherence, and painful editing workflows. That is typically the stage right before rails mature.

Next steps for video rails in 2026:

• Controllability: camera, style, structure

• Consistency: characters, products, brand rules

• Editability: local changes without regenerating everything

• Provenance: rights, disclosure, traceability

• Integration into creative workflows and distribution pipelines

Applications already emerged fast on the first video rails. But the winners will not be generic text-to-video tools. They will be end-to-end workflow products, such as performance marketing creative factories that ship variants at scale and learn what works.

The KPI becomes simple: speed to ship high-performing on-brand content and variants at scale.

3. New rails emerge on other data types: Tabular data is the sleeper

Text rails unlocked copilots first and autonomous agents now. Video rails unlock creation engines first and all-in-one content workflows now. The next big unlock comes when models become strong on other asset types: tabular data, scientific data, and eventually the physical world.

A near-term example I find compelling is AI for tabular data.

This applies foundation-model techniques to structured data in spreadsheets, databases, and warehouses to understand, predict, transform, and act on these tables. It matters because spreadsheets still run a huge part of enterprises, which gives tabular rails an unfair distribution.

If tabular data rails mature, here are the applications I expect:

• Automated cleaning, joining, and reconciling data

• Forecasting and scenario planning

• Anomaly detection

• Narrative reporting with traceability back to source tables

• Agentic spreadsheet workflows safe enough for finance and operations teams

This will look boring from the outside, but will be massive on the inside.

4. Multimodality unlocks AI for the physical world

A longer-term rail will be the so-called World Models. Intelligence is still captive in the digital world in any data type mentioned above, and breaking it out into the physical world will create a whole new range of use cases.

They will rely on multi-modality and will be able to combine vision, language, audio, and action.

Even though we are at the research stage here, I think early adoption in 2026/2027 will appear in some constrained environments with a clear ROI, such as:

• Logistics and warehouses

• Manufacturing

• Retail operations

• Medical care

• Domestic labor

My take is that voice, as the lowest-friction interface, will be crucial in unlocking real-time interaction, and video and 3D data will enable AI systems to go from perception to reasoning in the physical world.

The mental model for 2026

Rails mature → Applications emerge → Magic demos → Deceptive outcomes → Operational maturity → Scale → Authoritative capabilities.

The winners push through the disappointment phase. They solve edge cases, integrations, security and auditability, evaluation, and human-in-the-loop control.

2026 is not the year of better demos. It is the year rails become reliable enough to make applications unavoidable. Teams stop buying AI features and start buying systems that take responsibility for outcomes.

These are my 2026 bets, and I look forward to meeting the most ambitious founders tackling those!

When digital businesses go global, they collide with an anatomy of verification that’s deeply fragmented. From fluctuating identity registries to regional KYB workflows, from deepfaked documents to shifting regulations, the path to onboarding both individuals and businesses shouldn’t feel like navigating a maze. AiPrise solves that challenge with elegance: an orchestration layer that brings clarity, automation, and adaptability to global KYC/B. We believe AiPrise has the potential to transform how companies onboard users around the world.

Fragmented compliance is now a foundational problem

As companies scale across borders, regulatory complexity grows exponentially. What once could have been managed with a handful of vendors now demands a patchwork of integrations, each requiring onboarding, maintenance, and nuanced treatment per geography. The real friction is technical and strategic. Every moment spent juggling vendor relationships or patching together identity logic is a moment wasted building product and reach.

AiPrise abstracts away the complexity of compliance and reorganizes it into a coherent system. By orchestrating over 80 local and global vendors, the platform creates a single pane of glass for global onboarding. And orchestration is just the start. Dozens of financial services, marketplaces, remittance platforms, and embedded API-first businesses rely on AiPrise to route, fallback, and verify complex onboarding flows. These often involve both individuals and businesses across strongly regulated markets.

The opportunity is bigger than point solutions

Most identity providers focus on a single piece of the puzzle: a data API, document scanning, registry access, etc. AiPrise goes further. It owns the logic. If the match fails here, try there. If this registry is down, try a local provider. If the document is suspicious, trigger a manual review or AI check. It layers AI-based fraud detection on top of vendor verifications, blending document integrity logic, liveness, and synthetic detection alongside customer-selected providers. That combination of vendor flexibility and layered intelligence isn’t just different, it’s category-defining.

In many ways, AiPrise is building the operating system for global compliance. It doesn’t ask customers to shift to its stack, it adapts to theirs. And it does so in a highly modular, composable way. That opens the door for enterprise platforms and fintechs to embed AiPrise’s logic once and use it everywhere, everywhere meaning fully global.

Founders who deeply understand the problem

Chaitanya and his team have built AiPrise out of lived experience. They’ve seen firsthand how identity rules diverge across jurisdictions, how startups waste months onboarding vendors, and how fraud and compliance can derail global product expansion. That empathy shapes the product. Customers talk not about features, but about orchestration logic, rules-based fallback, and velocity delivered across 100+ countries.

There’s a strong team behind the vision. We’ve seen founders with global infrastructure ambitions before, but few who combine regulatory insight, execution discipline, and product empathy. The AiPrise team understands that regulation evolves unpredictably and that identity infrastructure cannot be brittle. It must adapt, fast.

Timing could not be better

A hundred years ago, journeys across empires required passports stamped at border crossings. Today, digital identities cross lines constantly. Companies are now launching globally from day one, and regulators everywhere from the EU to APAC to Africa are demanding better controls. Fraud and synthetic identity tools are more sophisticated than ever, and manual workflows struggle to keep pace. At the same time, cross-border payments are proliferating, with platforms moving money across dozens of markets daily. Each corridor comes with its own identity standards and compliance rules. AiPrise is uniquely positioned to simplify this complexity, giving companies a single layer to manage global onboarding while enabling the safe and efficient flow of payments worldwide.

Why this matters: a layer between product and regulation

At the end of the day, successful tech companies ship product, not compliance. But without compliance, you can’t scale globally. AiPrise solves that friction point by making regulatory infrastructure feel like plumbing - reliable, invisible, and infinitely scalable.

We’re thrilled to partner with AiPrise because they’re building the backbone for global onboarding and compliance. Their orchestration engine routes global identity and business verification flows with precision, clarity, and safety. They organize the flows, unlocking scale wherever business happens. And as companies increasingly default to global from day one, AiPrise ensures they can onboard with confidence anywhere.

From Scribe to AI Care Partner

Since our initial investment, Heidi has made the leap from being a best-in-class AI scribe to becoming a true partner in clinical workflows. They’ve introduced major new functionality across Calls, Context, Ask Heidi, Tasks, and Coding, and much more to come which, when woven together, represent the beginning of AI orchestration in healthcare. Doctors aren’t just using Heidi to record notes anymore; they’re leaning on Heidi to speed up tasks, manage follow-up calls for patients, support research, and ultimately operate as a system of action that makes their lives meaningfully better.

Growth That Speaks for Itself

The numbers underscore just how significant this shift has been. Since their launch 18 months ago, Heidi has supported more than 73 million patient consults and now supports over two million consults weekly in 110 languages across 116 countries. Their product-led strategy has scaled across multiple dimensions:

• Free to Paid: Rapid conversion proving tangible value to clinicians.

• Paid to Enterprise: Expansion into enterprise contracts across major healthcare systems from user champions has been proven. In the US they are working with hospitals like Beth Israel Lahey Health, the largest network of hospitals in Massachusetts and is cited as a “transformation partner” by MaineGeneral.

• Global Adoption: Active expansion beyond initial markets into new geographies, staying true to their “rest-of-world” strategy from day one.

Each of these steps reflects not just growth, but a repeatable go-to-market motion that positions Heidi as one of the fastest-scaling healthcare technology companies globally.

Why This Matters

Healthcare is facing unprecedented pressure. The World Health Organization projects a shortfall of 10 million healthcare workers by 2030. Research shows that clinicians spend nearly as much time on administration as patient care. Heidi is addressing this crisis by going beyond documentation to give time back to clinicians, reduce burnout, and improve the quality of care. In less than two years, they have already achieved scale and adoption levels that incumbents took decades to approach. Heidi has already returned more than 18 million hours back to clinicians.

Why We’re All-In, Again!

Our decision to reinvest was simple: Heidi continues to execute with precision, velocity, and vision. Their contrarian bets, prioritizing customization over clunky EHR integrations, going global from day one, and building a product-led engine out of Australia, have all paid off. And now, with AI Care Partner, they’re unlocking the next chapter of healthcare automation.

At Headline, we look for teams building generational companies that can fundamentally reshape industries. With Heidi, we see exactly that. The company is not only redefining what an AI healthcare assistant can do, but also showing the world what it means to put doctors and patients first. We couldn’t be more excited to continue this journey with them.

This week, Honeycomb launched Honeycomb Intelligence, an AI-native suite designed for exactly this future. With MCP Server, Canvas, and Anomaly Detection, Honeycomb is reimagining observability as something that doesn’t just live in dashboards and alerts, but directly inside the developer workflow with fast and interactive feedback loops.

MCP Server: Observability Meets the IDE

What it is: Honeycomb’s MCP Server connects AI coding assistants directly to telemetry data in Honeycomb. It’s now generally available with support for BubbleUp, heatmaps, and histograms.

How it works: At a basic level, MCP (Model Context Protocol) servers let AI assistants translate natural-language questions into data-backed queries. That’s table stakes. But Honeycomb has gone further: they’ve engineered an MCP that actually understands how to deliver the right context without overwhelming the model. Instead of feeding an AI agent an ocean of noisy telemetry, the Honeycomb MCP filters and structures the data so the agent sees exactly what it needs to reason effectively. This minimizes hallucinations, speeds up investigations, and keeps the AI focused on what matters.

Why it matters: Many vendors could ship a standard MCP server and call it a day. Honeycomb’s version is more compelling because it’s tuned for observability’s hardest problems: scale, cardinality, and context. It doesn’t just let an AI query your system — it ensures those queries are meaningful, performant, and actionable. That makes AI assistants not just functional, but trustworthy. It positions Honeycomb’s MCP as the connective tissue between development environments and production insight, where agents and humans can reliably debug issues without drowning in irrelevant data.

Canvas: A New Interface for Understanding Systems

What it is: Canvas is an AI-guided workspace that turns natural-language questions into multi-step investigations.

How it works: Engineers describe an issue — “Why did checkout latency spike yesterday?” — and Canvas automatically sequences relevant queries, surfaces charts, and highlights anomalies. Each step is shown visually (heatmaps, histograms, traces), and users can guide or redirect the AI. Canvas sessions can be shared with teammates or exported to Slack, where others can interact with the results directly.

Why it matters: The real power of Honeycomb’s approach to observability — unlike the traditional three-pillars model of logs, metrics, and traces — is its ability to give humans and AI access to rich, high-cardinality event data to uncover unanticipated connections and correlations. AI thrives on speed and context, both of which are enabled by Canvas. Instead of requiring query expertise or intimate system knowledge, engineers can simply ask questions in natural language and let the AI surface relationships that would otherwise stay hidden.

This transforms Canvas from a convenience into a force multiplier. It lowers the barrier for anyone on the team to uncover deep insights, it turns debugging into a collaborative process, and it unlocks the full potential of Honeycomb’s event-driven architecture by making complex system behaviors explainable in real time. In short, Canvas makes the strengths of Honeycomb’s data model broadly usable.

Anomaly Detection: Proactivity at Scale

What it is: An ML-driven engine that learns normal patterns across your services and surfaces deviations before they become incidents.

How it works: Honeycomb continuously monitors telemetry such as request rates, error counts, and latency, building baselines for each service. When actual behavior deviates significantly, the system raises an anomaly alert. Each alert includes context and integrates directly with BubbleUp and Canvas, so engineers can immediately see what’s different and why.

Why it matters: With hundreds of microservices and an explosion of telemetry, no team can hand-craft alerts for everything. Static thresholds can’t keep up. Anomaly Detection shifts observability from reactive firefighting to proactive protection. It reduces noise, spots regressions earlier, and ensures engineers aren’t blindsided by unknown unknowns.

The Bigger Picture: Why These Fit Together

Each of these releases is powerful on its own. But the real significance lies in how they work together as a unified vision:

• MCP Server ensures observability isn’t separate from development. It brings system insight into the IDE, and crucially, into the AI agents that are increasingly part of how code gets written. And unlike a basic MCP, Honeycomb’s implementation solves the hardest part — delivering the right amount of context so investigations are accurate, fast, and free from noise. Observability starts at the moment of creation, not after an incident.

• Canvas ensures those insights are usable and collaborative. It leverages Honeycomb’s unique strength — rich event-based telemetry capable of surfacing unanticipated connections — and makes it accessible through natural language and AI-guided investigations. Instead of requiring specialized expertise, Canvas turns the system into something that can explain itself, and makes those explanations easy to share.

• Anomaly Detection ensures no critical signals are missed. It provides the early warnings that feed into both MCP and Canvas, so engineers aren’t just asking questions after a failure. They’re investigating patterns proactively, guided by AI surfacing what’s abnormal.

Just as important is the foundation Honeycomb already built: a high-performance event-based observability platform capable of collecting massive volumes of high-cardinality data, correlating across traces, metrics, and logs, and delivering sub-second queries at scale. This rich telemetry and contextual depth is what makes Intelligence possible. Without clean, detailed event data and Honeycomb’s unique architecture, these AI-powered features would lack the signal to deliver useful insights.

Together, Honeycomb’s core observability engine plus Intelligence layer create a connected feedback loop: anomalies detected automatically, surfaced into developer workflows, investigated collaboratively, and continuously learned from. Code, telemetry, AI, and engineers form a closed loop, shortening the path from issue to insight and from insight to resolution.

Building Forward

The real potential of Honeycomb Intelligence isn’t in the individual features, it’s in what they unlock together. By tying real-time data, AI-guided investigation, and proactive detection into a single loop, Honeycomb is building toward something bigger: a world where systems explain themselves.

Imagine a future where every feature shipped comes with an automatic narrative of how it affected performance, where anomalies aren’t just flagged but contextualized in plain language, and where institutional knowledge compounds because every investigation becomes part of a living memory. Honeycomb is creating the conditions for engineering teams to move faster with more confidence, while learning from their systems continuously.

That’s the trajectory these releases point toward: observability as a compounding advantage, not just a safety net. Honeycomb isn’t just keeping pace with complexity — it’s redefining what it means to understand and trust the software you build.

Never in economic history have we witnessed a technology adoption curve as steep as AI’s. Smartphones took a decade to reshape consumer behavior. Social networks took years to reach critical mass. Generative AI, by contrast, has exploded into daily workflows in less than 24 months.

This is not just a matter of hype. It is measurable. Google’s own numbers tell the story: between April 2024 and April 2025, its token consumption increased 50x, reaching 480 trillion tokens per month. That is a proxy for real economic activity – every prompt, every query, every agent is another token consumed.

And it is only the beginning.

Growth at Unprecedented Scale

The leaders of this new economy are growing at a pace unseen in enterprise software. In the past six months alone, Anthropic has quadrupled its ARR to reach $4B. In the meantime, OpenAI, already larger, still managed to triple ARR to $12B.

What makes this wave unique is not only the size of the frontrunners but also the breadth of the ecosystem. Companies that did not exist five years ago now generate over $100M ARR: Mistral, Harvey, Abridge, Mercor, Lovable, Synthesia, Helsing, and more. In SaaS history, the journey to $100M ARR used to take a decade. AI-native startups are hitting the milestone in less than half the time.

The aggregate impact is already visible: AI generated $280B in revenue in 2024, a figure projected to reach nearly $2T by 2030.

Understanding the Waves of Adoption

Why has adoption been so fast? We see two reinforcing dynamics:

1. Technology meets ROI – As soon as technical advances become reliable enough for practical use cases, adoption spikes.

2. Talent shortages – In industries where skilled workers are scarce, AI tools are not optional; they are competitive advantages.

The result: penetration rates climb quickly, market leaders emerge faster, and growth compounds.

This dynamic has unfolded in identifiable “batches” of adoption:

• Batch 1: LLMs dealing with text and voice knowledge tasks have reshaped day-to-day jobs for many knowledge workers in different categories : Legal, Coding, Healthcare, Customer Support, Cyber.

• Batch 2: Image processing by AI through VLMs have created a new set of use cases impacting other categories such as Security, Marketing or Defense.

• Batch 3: With the agentic era, AI is starting to tackle increasingly complex workflows which enables enterprise integration in sectors like Finance, HR or Productivity.
  

• Next batch (our forward bet): Live decision-making and real-world support will have a deep impact at a broader scale. Think about AI-embedded robots already being built for Logistics use cases for instance. Or real-time decision making to trade on financial markets or balance the electric grid.

Each batch has produced category-defining companies. Each has turned AI into a tool that is augmenting humans, especially those in under-staffed sectors, rather than stealing people’s jobs.

Europe’s Right to Win

The US currently leads the AI race. They have nearly 70% of global compute power and the first AI behemoths originally came from the US. But Europe does have unique strengths that position it to produce the next global winners:

• Market Depth: 742 million people, nearly double the U.S. population.

• Talent Density: Home to world-class universities, research centers, and technical talent.

• Buyer Preference: European enterprises often favor local vendors for reasons of sovereignty, data control, and trust.

Europe is already proving itself. Companies like Mistral (France), ElevenLabs (Poland/UK), Helsing (Germany), Lovable (SWE), Synthesia (UK), and others have broken through.

At Headline, we believe the next generation of global leaders will also emerge here. That is why we have curated the AI Europe 100 – the most promising AI-native startups across all the categories above.

Headline’s AI Europe 100 - The Next Winners

The Road Ahead

The story of AI adoption is not a story of substitution. It is a story of amplification. Where human capacity is limited, AI multiplies it. That is why industries constrained by talent shortages are adopting AI fastest.

What is remarkable is that this amplification is happening at scale, across geographies and industries, and at a speed never seen before. From $280B in revenue today to nearly $2T within five years, AI is reshaping the world economy in real time.

The question is no longer if. It is where the winners will be built.

We believe many will be European.

Here is the full presentation from Headline’s AI event in Paris, presenting the AI Europe 100.

Some companies aren’t just better; they’re unrepeatable. They don’t fit a mold; they break the mold. These companies are “N of 1s”: rule-defying outliers that succeed not in spite of their singularity, but because of it.

N of 1 companies emerge at the intersection of the right product and the right cultural moment, but what makes them endure runs deeper. It’s a structural, emotional, or narrative edge that resists commoditization: something competitors can see but can’t replicate.

These consumer winners don’t live by formulas, frameworks, or spreadsheets. While SaaS companies might thrive by optimizing conversion rates and standardizing customer journeys, consumer category-creators are always rule-breakers, outliers in a market full of safe bets and copycats. Brilliance here isn’t found in a backtested model; it’s found in the exception to the rule.

Think: a magic notetaking tool that makes effortless a laborious-yet-essential task. A cult-like grocer daring to upend margin logic. A new vehicle that feels straight out of a sci-fi movie. A sleek ring that turns invisible health data into obsessive self-optimization.

To invest in consumer companies, you must hunt for rare magic: for companies that redefine entire categories through sheer force of vision. These aren’t just “better” products or “prettier” packaging. These are companies whose rule-breaking charm and unreplicable user experience truly separate them from seas of competitors building the same thing.

The market talks about “taste” as some vague superpower. At Headline, we focus on pattern-matching what makes N of 1 companies that consumers love, so we can recognize the next one before it becomes obvious.

The challenge of consumer investing is that by the time everyone can see the magic, it’s too late. The best N of 1 companies often look inexplicable in their early days: weird, niche, or serving needs that don’t yet feel universal. While there may never be a single formula, we wanted to unpack the elements we’ve noticed across the best ones that consistently separate the truly singular companies from the rest..

Companies’ N of 1 Superpowers

There are four clear categories of “n of 1” consumer companies, which are designated for their unique superpowers.

Product Edge

Some N of 1 companies win because the product itself feels like a breakthrough. They introduce a form factor or functionality that permanently shifts consumer expectations. Think of OŪRA, which turned a simple ring into a biometric tracker with cult adoption, or Granola, which eliminated the burden of note-taking entirely, delivering summaries so relevant it’s as if the AI knew exactly what mattered from your conversation. Once you’ve experienced that step-change, everything else feels like a step backwards.

Cult Following

Others build lasting advantage through mythology and identity. A24 created more than a film studio; it shaped a creative new worldview that fans proudly signal. Strava has become less of a run tracker and more of a social network, with athletes obsessively checking each other’s routes and performances. On TikTok, The RealReal fanatics trade tips for spotting mispriced items and post “graveyards” of pieces they regret not buying, turning secondhand luxury into a shared cultural game. Even Duolingo’s green owl has turned into a beloved character online, its memes extending the brand far beyond language learning. When a company becomes shorthand for belonging, its community naturally takes on the work of distribution.

Distribution Strategy

Every consumer company needs strong distribution to survive. The difference for N of 1s is that distribution is not just a necessity but the unfair edge that defines them. Skims mastered this by combining celebrity influence with algorithm-driven personalization, allowing the brand to live everywhere its customers already spend their attention. Netflix achieved the same by embedding itself into daily routines, using recommendations to make discovery feel effortless. For N of 1s, distribution evolves from survival mechanism to the very engine of dominance.

Pricing Advantage

Finally, certain N of 1s win by reshaping the economics of their category. Eight Sleep introduced subscription services to the category of sleep itself, turning what was once a one-time mattress purchase into ongoing sleep optimization. The Row has done the same in fashion, where $700 flip-flops can sell out because the brand’s stark minimalism makes dupes feel hollow by comparison. Their pricing power and consumers’ willingness to opt in reflect an aura that competitors cannot replicate. Whether through the invention of new revenue streams or the cultivation of brand-driven pricing power, these companies prove that price itself can be a moat.

N of 1 Companies and the Future of Consumer Culture

As we look ahead, we think the next wave of N of 1 companies will create cultural shifts rather than follow them. These founders will invent entirely new categories that feel inevitable only in hindsight.

In a world of endless competition and fleeting attention, “N of 1” consumer companies stand apart by building structural advantages, cultivating cult-like followings, and delivering products so compelling they become indispensable. These companies do more than win categories. They introduce entirely new ways of living in and seeing the world.

Part of this movement seems to stem from growing frustration with pre-COVID breakout vendors like Fivetran, Snowflake, Splunk, Auth0, Datadog, Confluent, and Snyk—many of which have since gone public, been acquired, or raised at premium late-stage valuations. As these companies shift focus toward their largest enterprise customers and profitability, many have raised prices while simultaneously slowing product velocity. Buyers, in turn, are pushing back against their high spend on these platforms, which is becoming increasingly hard to justify.

There is now a real opening for challengers offering simpler, leaner, and more cost-aligned alternatives. The 10 trends below explore where we see the infrastructure software market headed.

Key Trends

1. Market Preference Is Shifting Back to Best-of-Suite Platforms

Buyer sentiment is shifting back toward best-of-suite solutions, as teams grapple with tool sprawl and vendor fatigue following the era of best-of-breed adoption. The trend is especially evident in security, where Gartner found that 75% of organizations are actively pursuing vendor consolidation. Major M&A activity reflects this shift—two of the largest deals of 2025 are security consolidation plays: Google’s $32B acquisition of Wiz and Palo Alto’s $25B acquisition of CyberArk.

The same dynamic is playing out among startups, where several challengers are seeing early success by centering their value proposition around a best-of-suite approach.

Examples:

• Networking: NetBox Labs — Composable platform for managing network infrastructure, including operations, observability, automation, and security

• Cybersecurity: Aikido Security — Comprehensive security platform spanning AppSec and cloud security

• DevOps/SRE: Incident — All-in-one incident management platform covering alerting and incident response

2. Transparent Pricing Is Becoming a Differentiator

As incumbent pricing grows increasingly opaque, buyers are migrating to challengers with clearer, more predictable pricing models. In many cases, renewal shock—driven by quietly compounding usage or gated features—pushes teams to reevaluate long-standing vendors.

Examples:

• Observability: Datadog → Honeycomb — Honeycomb prices based on data ingest—but in atomic units which can contain an effectively unlimited number of key/value pairs (sometimes called “custom metrics”) tied to a given timestamp. This allows Honeycomb pricing to be more predictable, as costs are simply based on request volumes rather than a mix (or layering) of per-host, per-product, or per-attribute costs.

• ETL/ELT: Fivetran → Estuary — Fivetran charges based on Monthly Active Rows (MAR)—the number of rows inserted, updated, or deleted each month—which can spike unpredictably in high-churn datasets. Estuary avoids this with transparent pricing based on data volume and connector count.

• AppSec: Snyk → Aikido Security — Aikido charges a flat per-user rate that covers its full suite of modules across AppSec and cloud security—avoiding the per-seat plus per-feature escalation common with Snyk’s licensing model.

• Auth: Auth0 → Clerk — Clerk’s pay-as-you-grow model avoids Auth0’s MAU caps and tiered gating, keeping pricing predictable while advanced features remain add-ons rather than Enterprise-only.

• Business Intelligence: Looker → Omni — Omni offers a single, predictable monthly rate with no hidden fees or per-role pricing. Looker, in contrast, frequently charges separately for dev, standard, and viewer licenses and ties costs to compute usage and feature access—resulting in higher and less predictable bills.

3. The Modern Data Stack Is Converging on a Leaner, Modular Core

As complexity and cost have accumulated in the modern data stack, many teams are converging on simpler, more modular foundations. Rather than over-centralizing around warehouses or over-engineering pipelines, teams are relying on relational databases and object storage to do more—each extended just enough to support evolving needs without bloating the stack.

Relational databases are becoming increasingly versatile. New vendors and extensions are enabling teams to extend their existing databases to support functionality that has historically required multiple systems. For example, PuppyGraph, pgvector, and ParadeDB minimize the need to run separate graph, vector, and search databases, respectively. Reducing the number of systems that teams need to purchase, operate, sync, monitor, and secure meaningfully lowers administrative overhead, cost, operational burden, potential downtime, and the surface area for security breaches. Postgres’s rich extension ecosystem has made it a prime beneficiary of this shift, fueling its continued rise in popularity.

Object storage is having a similar moment, becoming the foundation for a growing range of workloads in analytics, search, and data processing, where cost, scalability, simplicity, and durability take precedence over fully optimized performance. In analytics, engines like Trino and even ClickHouse can query object stores directly using open table formats such as Apache Iceberg and Delta Lake. In search, AWS recently introduced S3 Vectors as the first cloud object store with native vector storage and querying. turbopuffer extends this approach by enabling both vector and full-text search on object storage, supported by a caching layer to boost performance. Further upstream, Bauplan aligns with the trend by offering serverless abstractions for building data pipelines, often for AI/ML use cases, directly on top of object storage.

Together, these trends reflect a broader move toward leaner infrastructure: fewer systems to manage, fewer pipelines to maintain, and more leverage from modular, general-purpose, and cost-effective primitives that can flex to support modern operational, analytical, and AI workloads.

4. Cost Optimization Platforms Are Emerging Across the Stack

As infrastructure costs become a board-level concern, a new class of tooling is emerging to reduce spend across compute, storage, and software. While Kubernetes and cloud billing were early focus areas, nearly every major cost center now has dedicated solutions, including spaces like data warehousing and business intelligence. Some tools, such as Pump, take a fintech-style approach, leveraging group buying across customers to obtain volume discounts and unlock the benefits of reserved instances. Others, such as Espresso AI, focus on optimizing utilization and footprint—identifying overprovisioned workloads, inefficient queries, or unused assets to reduce waste at the source.

Examples:

• Cloud Billing: Pump — Aggregates usage, unlocks preferred pricing on reserved instances, and simplifies cloud cost visibility

• Data Warehouses: Espresso AI — Surfaces inefficient queries, redundant schedules, and unused tables to shrink compute/storage spend

• BI Tools: Datalogz — Detects stale dashboards, orphaned models, and unused reports to reduce unnecessary BI layer costs

5. A New Class of Vendors Is Winning on Cost-Aware Design

A new class of infrastructure vendors is emerging—distinct from traditional cost optimization platforms—that targets high-cost line items like CI compute and SIEM. Rather than positioning solely as cost cutters, these players offer optimized, drop-in components that improve performance and efficiency while reducing compute and storage overhead as a natural byproduct. In many cases, they are replacing brute-force, overprovisioned systems with leaner, AI-native primitives that become embedded parts of the modern stack. Cribl pioneered this positioning, and a new wave of entrants is now applying the same playbook across the infrastructure stack, often with AI-native positioning.

Examples:

• CI Compute: Namespace — Provides faster, lower-cost CI runners—combining proprietary hardware with intelligent caching to serve as a drop-in replacement for managed runners across platforms like GitHub Actions

• Security Data: Realm Security — Lowers SIEM costs (e.g., from Splunk, Sumo Logic) by acting as an upstream telemetry broker—intelligently filtering and routing high-volume, low-signal data to cheaper destinations

6. Challengers Are Capturing the Underserved Mid-Market

While much of the narrative discussed in this piece thus far centers on migration from legacy platforms, a parallel trend is unfolding: emerging vendors are capturing greenfield adoption by targeting the historically underserved mid-market. Incumbent solutions are often prohibitively expensive, require significant customization, and demand heavy implementation and maintenance resources—creating friction for leaner teams with constrained budgets. In contrast, new entrants are prioritizing simplicity, accessibility, and rapid time-to-value, delivering solutions that “just work” for mid-market buyers without the overhead of enterprise tooling. As part of this shift, cloud-native vendors are also introducing low-friction, modern alternatives that replace the complexity of clunky on-prem incumbents.

Examples:

• Data Modeling: SqlDBM — A cloud-native alternative to Erwin

• Cybersecurity: Aikido — For teams where managing the Snyk + Semgrep + Wiz stack is excessive

• Artifact Repositories: Cloudsmith – A cloud-native alternative to JFrog Artifactory

7. AI-Native Service Delivery Is Transforming Outsourcing Models

Most companies—especially outside the Global 2000—lack a compelling reason to manage IT, security, or QA functions internally. While outsourcing has long been an option, AI is materially shifting the cost–quality frontier. Service providers can now deliver faster, more consistent, and more scalable outcomes at equal or lower price points. As a result, we expect a growing portion of these functions to be externalized over time. This shift is playing out through both AI-native service providers that deliver outcomes directly and AI-native software platforms that enable existing providers to automate more and operate at higher margins.

Examples:

• AI-native MSP software: Thread

• AI-native QA service: QA Wolf

• AI-native MDR services: TENEX

8. Open Source Alternatives Are Gaining Favor

Open source tools offer a practical alternative for teams with the capability to self-manage infrastructure. By avoiding the high costs of managed hosting and usage-based pricing, OSS provides greater cost control and data ownership while reducing vendor lock-in. As infrastructure costs rise and scrutiny around SaaS sprawl and data privacy intensifies, more teams with in-house expertise are expected to adopt open source as a preferred path. For example, as Postman has matured and raised pricing, many companies have found it difficult to justify the spend versus emerging OSS alternatives like Bruno. Even in more frontier areas, OpenAI’s release of GPT OSS—an open-weight model suite—signals the growing importance of openness in AI.

Examples:

• Observability: Datadog → Prometheus + Grafana

• ETL: Fivetran → Airbyte

• API Testing: Postman → Bruno

9. Vendor-Native Tooling Is Quietly Becoming the Default Path

In parallel with the shift back toward “best of suite,” we are seeing increasing adoption of vendor-native tooling—particularly from hyperscalers. In security, the rapid growth of Google SecOps and Microsoft Defender underscores how bundling continues to gain traction. With executive teams under pressure to rein in budgets, many are steering functional groups toward platform-native solutions, especially when they have pre-committed platform spend. As owners of the underlying infrastructure, hyperscalers can squarely outcompete other vendors on cost

A similar dynamic is unfolding in AI, where OpenAI and Anthropic are actively moving up the value chain into the application layer. Anthropic’s Claude Code and OpenAI’s Codex are examples of this shift in software development, while OpenAI’s investment in Endex signals broader ambitions to expand into verticals such as financial services. Just as hyperscalers benefit from owning infrastructure, OpenAI and Anthropic enjoy a strategic advantage by owning the underlying models.

Examples:

• Security: Choosing Microsoft Sentinel, Google Chronicle over Splunk

• DevOps: Choosing AWS CloudFormation over Terraform

• AI: Choosing Claude Code over Cursor

10. There Is Still Appetite to Pay for Clean Abstractions of Hard Problems

Despite tighter budgets, teams remain willing to pay for clean, reliable abstractions of foundational but operationally complex problems—especially those at the intersection of application logic and infrastructure complexity. In many cases, the cost of replicating these capabilities internally exceeds the price of adoption when accounting for headcount, opportunity cost, and ongoing maintenance. Even in lean environments, the tradeoff often favors outsourcing over in-house build-and-maintain efforts. For example, replicating Recall’s functionality in-house could require a team of 5–6 engineers working for several months, followed by ongoing operational overhead—making its API abstraction a far more cost-effective path for teams looking to move quickly and leanly.

This dynamic—coupled with the rapid growth of agentic applications—is powering the rise of a new category of API-based, AI-native abstractions purpose-built to support these emerging systems. Just as Stripe abstracted payments and Twilio simplified messaging, platforms like Composio, Browserbase, Daytona, Tavily, and Zep offer simple, clean interfaces for handling auth, integrations, runtime environments, web search, and context management. These abstractions are becoming the default path for teams looking to offload complexity, maintain development velocity, and build reliable, production-grade agent workflows without reinventing core infrastructure.

Examples:

• Meeting bots & recorders: Recall

• Auth & tools: Composio

• Headless browsers: Browserbase

• Code sandboxes: Daytona

• Web search: Tavily

• Context: Zep

Closing Thoughts

The infrastructure software landscape is undergoing a meaningful reset. As AI-native solutions command attention, buyers are simultaneously reassessing the traditional infra stack—seeking solutions that are simpler, leaner, and more cost-effective. The trends we highlight reflect a broader shift in mindset: away from sprawling, over-engineered systems and toward software shaped by pragmatic design.

The bar for infrastructure products has never been higher. Buyers are increasingly discerning—ruthlessly evaluating tools based on performance, ROI, and total cost of ownership. We expect a range of models to win in this new era: from next-gen platforms and open core products to AI-enabled services, clean abstractions, and systems that meaningfully reduce operational burden or cost. The next generation of infrastructure will be defined by products that are pragmatic, performant, and deeply aligned with how modern teams build.

As always, if you’re building anywhere in the infrastructure stack, please feel free to reach out to jacob@headline.com to connect.

That’s the world Citizen Health set out to change.

Today, Headline joins a $30 million Series A round alongside 8VC and prior investors Transformation Capital, Wavemaker 360, and the Chan Zuckerberg Initiative (CZI)—all of whom bring deep expertise and networks in healthcare and technology.

At its core, Citizen Health is building a beautiful consumer platform that brings together every piece of a patient’s health journey—clinical data, genetic information, imaging, patient-reported outcomes, and claims. They’re then using AI to make sense of all this data, turning a pile of disconnected records into a clear, actionable story for each patient. For families, this means having a digital interface where they can actually find and use their medical information, get matched to clinical trials, make sense of medications and treatments, and connect with others who understand what they’re going through.

For healthy consumers, this interface may not sound very critical but for rare disease patients and their families, these records are typically inaccessible or scattered across numerous portals, screenshots, and printouts and can be the difference of understanding whether they qualify for a new treatment or will have an adverse reaction to a drug. On Citizen, all records can be transferred to a healthcare provider with the click of a button.

But the impact goes way beyond individual patients. By partnering with patient advocacy groups (PAGs), Citizen Health is also solving a massive data fragmentation problem. There are about 1,200 PAGs in the U.S., each with its own registry and approach, which unintentionally keeps valuable data siloed. Citizen Health’s platform acts as a universal connector, aggregating data across all these groups, making it far more useful—not just for patients, but for the researchers and companies racing to develop new therapies.

Citizen Health was spun out of Invitae, bringing a powerhouse team with deep roots in both technology and patient advocacy. CEO Farid Vij, who was previously COO at Ciitizen (acquired by Invitae for $325M), and Chief Business Officer Nasha Fitter, a widely respected rare disease advocate, helped Citizen Health to build early trust with patient advocacy groups and families. From a business perspective, Citizen’s approach is already paying off. Since its seed round, Citizen Health has brought on dozens of PAGs and thousands of patients. Nearly half of their customers have expanded their business with Citizen Health, either by adding new disease cohorts or new use cases. Big names like AstraZeneca, UCB, and Ionis are early customers.

In addition to the impact on patient lives, we are very excited about the flywheel that Citizen Health is starting to see. Historically, biotech and pharma companies have struggled to develop treatments for rare diseases because of the lack of available patient data. As more PAGs and patients join the platform, their data becomes richer and more valuable. This attracts more enterprise customers and drives more revenue, enabling Citizen Health to invest further in their technology and outreach. This, in turn, draws in even more PAGs and patients, creating a virtuous cycle of growth and impact.

Citizen Health’s growth has been organic, driven predominantly through word-of-mouth due to Nasha’s advocacy for FOXG1 and credibility in the rare disease space. As AI continues to drive down the data structuring costs, their B2B2C model keeps customer acquisition costs low.

Unique in healthcare, the consensus among patients, advocacy leaders, and pharma clients was glowing and consistent during our diligence process. PAG leaders raved about how Citizen finally gave their communities a way to organize and use their health data. Pharma clients highlighted the unique value of Citizen’s data across their efforts in new drug development, trials, and even patient recruitment. And everyone—from customers to competitors—spoke highly of Farid, Nasha, and the entire Citizen Health team.

Through their own experiences, Citizen deeply understood that rare disease patients never have beautiful products built for them – constantly told that the market opportunity isn’t great enough for great innovation. We admire that the team sees patients as consumers, who deserve intuitive and modern interfaces to navigate a notoriously complex system. We hope that in the future, when anyone is sick, the first question their doctor asks is, “Are you on Citizen?”

Citizen Health is the perfect combination of the right team + right technology + right approach + right time, colliding to solve one of healthcare’s most challenging problems. Citizen Health isn’t just building a business, but creating a platform that could redefine how we think about managing rare diseases–and beyond.

For more information, visit citizen.health.

As investors, we look for transformational platforms built on deep domain understanding, with an opportunity to own workflows and data in a large, underserved market. Thread checks all the boxes. Its AI-native platform is redefining how MSPs triage, assign, resolve, and automate support – not incrementally, but fundamentally.

Here’s why we believe Thread has the potential to become a generational company.

Why This Market? Why Now?

The managed services market is enormous: Over 150,000 MSPs globally serve millions of SMBs and mid-market companies. Yet despite their critical role, most MSPs still rely on email inboxes, spreadsheets, and billing systems to handle service and support.

Thread enters at a perfect moment. MSPs are under pressure to do more with less: clients expect real-time, always-on support, but technician headcount can’t scale linearly. Meanwhile, AI is unlocking 10x gains in efficiency across organizations. From GitHub Copilot in development, to AI playbooks in sales, to autonomous chatbots in customer support - the future of work is AI-augmented.

Service delivery is next. Thread is building the AI-powered foundation for that shift.

Thread’s AI Service Desk: A Full-System Redesign

Thread replaces legacy helpdesk infrastructure with a purpose-built AI platform. Its flagship product is a fully integrated service desk with:

• Magic Agents: AI assistants that greet end users instantly and autonomously handle routine support requests. These agents triage issues, collect structured data, trigger automations, and resolve tickets – often without human intervention.

• Automated workflows: Thread connects to PSAs, RMMs, documentation systems, and RPA platforms to orchestrate resolution end-to-end. No more copy-pasting between tools.

• Intelligent technician tooling: For escalated tickets, Thread provides technicians with complete context, AI-generated summaries, resolution suggestions, and even automated closing notes and time entries.

The result is support that’s faster, cheaper, and better. MSPs using Thread have reported:

• 5x faster responses;

• 78% faster resolution times;

• 86% of tickets resolved same-day;

• 300+ hours saved per month in manual triage, logging, and dispatch.

Value Flywheels: AI Agents + Workflow Ownership + Data

What excites us most is what lies beneath Thread’s UX:

1. Autonomous Service Agents: These aren’t chatbots. Thread’s agents operate like AI coworkers, gathering info, triggering fixes, updating systems, and closing loops. This isn’t assistive AI; it’s AI doing the work.

2. Workflow Depth: Thread integrates across the MSP stack – from ConnectWise and Autotask to RPA and internal documentation. It doesn’t ask MSPs to rip and replace; it meets them where they are, and gradually absorbs more of the workflow.

3. Data Ownership: Every ticket, resolution, and automation flow becomes part of a growing intelligence layer. Over time, Thread will build the most complete dataset on service delivery across MSPs. That feedback loop makes its agents smarter and more useful with each deployment.

In a fragmented industry underserved by modern tools, Thread is positioned to become the default AI operating system for MSPs.

Why We Backed the Team

Thread’s co-founders, Mark Alayev and Matt Linn, experienced the MSP pain firsthand. They both spent time at an MSP, where they rose through the ranks by innovating and operationalizing the business to over 300 employees. In leading that service provider, they learned exactly how broken service delivery is - along with how to fix it. They’ve since assembled a technical and go-to-market team with deep empathy for their customer and a track record of shipping fast.

In a short amount of time, they’ve built a world-class product, integrated with the core MSP stack, attracted tremendous new leadership talent, and deployed with more than 500 MSPs serving over 10,000 small and medium businesses worldwide. The traction is real, their customers are thrilled, and the Thread vision is far bigger than tickets and chatbots. We have no doubt that Mark, Matt, and new executive Michael Evers make up a team uniquely capable of executing on that vision.

TL;DR

Thread is an AI-first, deeply integrated system of action for the future of service. We believe every vertical will have an AI-powered operating system. For MSPs, that will be Thread.

We’re honored to be early partners.

As with last year’s inaugural list, we compiled a set of the most compelling companies based on submissions and conversations with founders and investors. We then further narrowed down the candidates by stage (pre-seed through Series A), geography (North America), funding history (less than $25M raised) and momentum.

Over the past three years, vertical software has undergone a massive evolution—none more dramatic than what we’re seeing today with the rise of vertical AI. We’re witnessing unprecedented growth across industries, including those historically resistant to digital transformation. At Avid, we’ve seen this firsthand with companies like Basis - at Headline, it’s Heidi Health and at BCV, Lightsource is leading the way.

The first wave of AI tools found traction in desked industries like legal, where knowledge work could be more easily abstracted. But what’s most exciting about this year’s VSaaS 50 is the emergence of AI-powered software in deskless, high-touch industries—utilities, energy, biotech, supply chain, manufacturing, healthcare, and beyond. These are sectors where software adoption has traditionally been slow, yet the impact of AI + vertical SaaS is proving transformative.

As software becomes cheaper and more accessible, we believe the most valuable platforms will win by driving industry-specific workflows, vibrant user communities, and world-class GTM execution. This is a once-in-a-generation opportunity to serve industries hungry for modernization—by earning their trust through thoughtful, purpose-built products.

This year’s Vertical SaaS 50 highlights the companies we believe are poised to become category-defining leaders in the industries that need them most. If you’re building the next generation of Vertical SaaS and Vertical AI solutions please email taylor@headline.com, zeeza@avidventures.com or cmelaskyriazi@baincapital.com, we’d love to chat.

Written by Taylor Brandt, Partner at Headline, in collaboration with Zeeza Cole, Partner at Avid Ventures.