The AI-native economy

Build, launch, & grow
autonomous startups.

An autonomous startup is the whole company — written as Business-as-Code, then sold as Services-as-Software: your customers buy the outcome, not the seats.

Why now

Every era builds its companies the way it builds its software.

In the cloud era, Infrastructure-as-Code turned servers into software — and that’s what unlocked Software-as-a-Service. The same shift is happening one layer up: Business-as-Code turns the company itself into software — and that unlocks Services-as-Software, where you don’t buy a tool to do the work, you buy the work done.

The $50-trillion services economy is being rewritten. The only question is which side of it you build on.

Cloud eraInfrastructure-as-CodeSoftware-as-a-Service
AI eraBusiness-as-CodeServices-as-Software

IaC : SaaS  ::  BaC : SaS

Our method

Business-as-Code.

Everything that makes a company — who it serves, what it sells, how it sells — written as code. Versioned, forkable, and executable, so a startup is something you ship, not something you staff up.

acme/business.ts
import { Business, agent } from 'business-as-code'

export const acme = Business({
  name: 'Acme Books',
  mission: 'Close the books for every small firm — automatically.',
  icp: 'US accounting firms, 2–200 staff',
  goals: ['500 firms onboarded', '$1M ARR', '99.9% on-time filings'],

  // sold as an outcome, not seats — software margins on a services budget
  products: [
    { name: 'Monthly Close', model: 'subscription', price: 499, cogs: 40 }, // ~92% margin
  ],

  // staffed by named digital employees — one per function
  team: [agent('Sally', 'Sales'), agent('Finn', 'Finance'), agent('Tom', 'Tech')],

  metrics: ['mrr', 'churn', 'grossMargin', 'closeTime'],
})

Autonomous where it’s safe. Human where it matters. Services-as-Software isn’t about replacing people — it’s about freeing them from execution to do the work only people can do.

The category

Buy the outcome.

SaaS sold you a tool and a login, then billed per seat while you did the work. Services-as-Software sells the work finished — the return filed, the contract reviewed, the books closed. Software’s margins on labor’s budgets.

Traditional services10–30% margin
Services-as-Software85–95% margin

AI can already do this work — most of the services economy just hasn’t bought it that way yet. That gap, between what’s possible and what’s been adopted, is the widest it will ever be.

AI capabilities have outrun adoptionTwo exponential curves on a shared timeline. AI Capabilities is already vertical at the left edge; AI Adoption traces a long flat tail before turning steep at the right. The horizontal gap between them is the Services-as-Software opportunity.
AI Capabilities
AI Adoption
The Services-as-Software opportunity

The loop

The market is the reward function.

The agents don’t just run the company — they optimize it. Every startup is a live experiment: they test ad creative, landing-page copy, pricing, and offers, and the market grades them in real numbers. Because the business runs on real payment rails, that reward is ground truth, not a proxy. Winners compound, losers get cut — and the company improves while you sleep.

01Ad CTR
02Signup %
03Activation
04Revenue
05Growth
06Profit

proxies    reward · the agents optimize the whole funnel

What we build

Three first-class categories.

Every autonomous startup is one of three: an Agent, a Service-as-Software, or a Software. Each runs as Business-as-Code.

Crossed with every business model — B2B, B2C, B2A, marketplace, API, data, directory — and every niche in the economy. Millions of possible Startups, generated on demand. A curated Canonical Index keeps only what’s genuinely distinct — so the atlas stays a market map, not spam.

The engine

From a coordinate to a company.

This is how we build. The Combinator takes any niche in the economy and runs it through a cascade of typed stages. Out the other side: a live, selling startup.

  1. emit
    Coordinate

    A cell in the economy's grid — the niche, addressed.

  2. emit
    Problem

    The pain worth paying to remove, stated from the operator's chair.

  3. split
    ICP

    Who exactly buys — one profile per branch.

  4. split
    Solution

    What ships for that profile — scoped, never generic.

  5. split
    Differentiator

    Why this entrant wins its niche.

  6. bind
    Founding Hypothesis

    The bet, written down and testable.

  7. bind
    Position

    The claim staked against the market.

  8. bind
    Offer

    What the customer buys, priced as the outcome.

  9. bind
    Business-as-Code

    The whole company, written as a program.

  10. bind
    Financial Rails

    Commerce wired in from the first commit.

  11. emit
    Agents

    The team, hired from the roster — named digital employees.

  12. emit
    Surfaces

    Site, API, docs — every door the market walks through.

  13. runtime
    Live startup

    The cascade’s output — running, selling, compounding.

Who it's for

Build it, or fund it.

Builders & founders

Build a startup.

Technical or not — you bring the industry knowledge; the engine and its agents do the building and the staffing. Start your own or join a portfolio company.

Investors

Fund a portfolio.

Two ways in: apply for dealflow from the companies we launch and back them directly — or hold the whole portfolio in one position as an LP in the Studio Fund.

Come build with us.

If you’re an ambitious founder with unique insight into a niche, the studio was built for you — we handle the technology while you focus on the business.