Pick a service. There's already an AI-native version of it.
by Ygor Fonseca, Founder & Systems Lead
In one quarter, the AI-native services category went from a thesis to a roster. Six verticals. Six named companies. Six valuations totalling north of $20 billion. And in every one of them, the same shape underneath: AI does the reading and the synthesis, an accountable human owns the judgment, and the service is sold productized — not billed by the hour.
Marketing is one of the last big-budget services without a fully formed AI-native option in the small-and-mid-market band. That gap is what Leanboat is. But before getting to that, look at what is already true in the categories that beat marketing to the punch.
The six
Insurance brokerage. Harper rebuilt the broker model and reported 30x revenue per rep in their Series A announcement earlier this year. The mechanic: AI reads the policies, runs the comparisons, prepares the recommendations; the licensed broker holds the customer relationship and the final advice. A team of 30 does the work that used to need 300.
Legal contracts. Crosby raised a $60 million Series B at $400 million valuation on the back of a single number: 58-minute median turnaround on contract review across 13,000 contracts. The pricing change is sharper than the speed change — Crosby sells contracts at fixed prices per document, not at billable hours. Senior attorneys hold judgment. Agents handle the reading.
Accounting. Basis raised a $100 million Series B at $1.15 billion and is now running tax and audit workflows for roughly 30% of the Top 25 US accounting firms. Same shape: agents read the documents, accountants own the judgment, the firm sells outcome instead of hours.
Legal research. Harvey reached $11 billion in valuation — the largest AI-native services valuation to date. Their product is read-and-synthesize at scale across case law, contracts, and corporate documents, with attorneys holding the final answer.
Recruiting. Mercor quintupled to $10 billion in valuation in a single year by rebuilding the talent-matching layer with AI on top of human evaluation. The shape repeats. The platform reads at scale, the recruiter validates the match, and the company is paid on placement, not on time.
Residential real estate. Modern Realty (YC S24) is the youngest of the six — and the most direct rebuild of a service category most people assumed could not be productized. Same architecture as the others: agents handle the comparable-sales analysis, the title-and-disclosure read, the negotiation prep. Licensed agents hold the client relationship. Buyer rebates replace inflated commissions.
The shape that connects them
If you look at the six side by side, the architectural pattern is so consistent it stops feeling like a coincidence and starts feeling like a category definition.
AI does the reading and the synthesis. Every one of these services is heavy on documents, comparables, prior cases, structured data — the work that takes a junior or mid-level operator hours and that AI does in minutes. This is the layer where the productivity gain shows up.
An accountable human owns the judgment. Every one of these companies has a licensed, certified, or named operator at the interface with the customer. The agent does not file the brief, the lawyer does. The agent does not bind the policy, the broker does. The agent does not close the deal, the realtor does. The judgment is what the customer is buying. The AI is what makes the judgment cheap to deliver.
The service is sold productized, not by the hour. Crosby charges per contract. Harper charges per policy. Basis charges per workflow. The unit price is anchored to a deliverable, not to the time it took to produce. This is the pricing change that makes the AI-native economics actually work — because if you bill by the hour and AI cuts your hours in half, you cut your own revenue in half.
Sequoia's Inference newsletter put the pricing thesis directly: the unit of value is the decision delivered, not the time spent producing it. The companies above are early proof.
Why marketing is the lagging vertical
Marketing has every condition the six categories above have. It is document-heavy (briefs, plans, reports, creative assets, customer research). It rewards judgment that an accountable operator can hold. It is one of the largest single budget lines in any consumer or software brand. And the legacy delivery model is still mostly hours-against-retainers — the same shape that legal, accounting, and recruiting all just rebuilt.
What is missing is not the technology. The agents work. The integrations work. The data is queryable. What is missing is a small enough number of operator-led companies actually shipping the architecture in marketing — productized service, AI doing the reading, human holding the judgment, priced against outcome rather than hours. The handful that exist are mostly above the price band where small and mid-market consumer brands actually buy.
That gap closes one of two ways. The legacy agencies productize down — but most cannot, because their cost structures are built on the hour. Or new operators build the AI-native version directly, at the price band where the buyer actually lives.
The second is what Leanboat is. An AI-native agency for consumer and software brands at the scale where most growth budgets actually sit. Same architecture as Harper or Crosby, applied to the marketing function: AI does the production and the analysis, operators hold the judgment and the customer relationship, and the service is sold against the outcome it ships.
If you spend any meaningful share of your budget on marketing services, the question to ask the vendor on your next renewal is the same one a CFO would ask Basis or a GC would ask Crosby: what is AI actually doing in the delivery, what is the human doing, and what is the price tied to? That is the test. The answer separates AI-painted from AI-native.