Being built at the Dealroom hackathon · Thu 1 Oct 2026 · London
How much company does each employee buy?
On Thursday 1 October we're joining The Autopilot Index, a free, evidence-graded index of companies that say AI runs them, to Dealroom's company data at the Dealroom hackathon. The goal is one screen: funding per employee and revenue per employee, set against how much of the business AI actually runs (L2 to L5). Until then, this page shows our own index data, and every figure on it has a source and a date.
30 companies tracked · 2 on audited or filed evidence · 6 on founder claims alone
The screener
What the screener does
Joins two datasets
We're matching each company in The Autopilot Index (autonomy level, evidence grade, reported humans, dated financial observations) by domain to Dealroom's records for funding rounds and headcount.
Computes leverage
Funding per employee and revenue per employee, each shown with the date and source of both the numerator and the denominator. If either side is missing or undated, the ratio stays blank. It is never estimated.
Ranks it against autonomy
Companies sit in bands by their published autonomy level (L2 to L5). L1 Copilot is out of scope: the Index only covers businesses where AI does the work, not tools that help a human go faster.
Nothing on this page comes from Dealroom. The leaderboard below uses only The Autopilot Index's own data.
Ask the index from your agent
Our public MCP server serves the index data on this page through read-only tools (search_companies, get_company, list_stack_tools, list_editions). No login.
{ "mcpServers": { "autopilot-index": { "type": "http", "url": "https://autopilotindex.com/mcp" } } }More in the developer docs.
What the levels mean
| Level | Name | Definition |
|---|---|---|
| L1 | Copilot | Out of scope. AI assists; humans operate |
| L2 | Function autopilot | AI runs whole functions end-to-end (support, growth, build) with human review |
| L3 | Operational autopilot | AI runs most core operations; human steers the big calls |
| L4 | Goal-level autopilot | Human sets goals & signs papers; AI operates the company day to day |
| L5 | Full autonomy | AI runs the company including capital allocation |
Autonomy levels are editorial assessments, not audits. Evidence grades record source strength, not proof of autonomous operation. Evidence grades: A Audited or independently reviewed financials · B Public transaction or filing · C Press coverage with founder confirmation · D Founder claims only.
Leaderboard
The index leaderboard
Index companies only (not the watchlist), ordered by evidence first, then autonomy. Our own dated, source-linked data; a blank cell means not disclosed.
| Company | Level | Evidence | Reported humans | Latest financial observation | Total raised |
|---|---|---|---|---|---|
| Medvi | L2 Function autopilot | Grade AAudited or independently reviewed financials | 2Observation date unknown · unclassifiedNYT (financials reviewed) | $401M sales in 2025Revenue · 2025-01-01 to 2025-12-31NYT (financials reviewed) | $0 — bootstrapped and profitableDate not recorded |
| Base44 | L2 Function autopilot | Grade BPublic transaction or filing | 1 founder + 8 employeesAs of 2025-06-18TechCrunch (1 founder + 8 staff) | ~$150M ARR under WixARR · As of 2026-05Wix Q1 2026 results | $0 (bootstrapped)As of 2025-06TechCrunch |
| Boardy | L3 Operational autopilot | Grade CPress coverage with founder confirmation | not disclosed | not disclosed | $11MAs of 2025-01TechCrunch |
| Polsia | L3 Operational autopilot | Grade CPress coverage with founder confirmation | 1 founder; zero employees reportedAs of 2026-05Pulse2 · self-reported | Approaching $10M annual run rateAnnual run rate · As of 2026-05Pulse2 · self-reported | $30MAs of 2026-05Pulse2 |
| Gamma | L2 Function autopilot | Grade CPress coverage with founder confirmation | 50 employees (founder announcement)As of 2025-11-10Gamma founder announcement | More than $100M ARRARR · As of 2025-11-10Gamma founder announcement | ~$90M (incl. secondary)As of 2025-11TechCrunch |
| Nanocorp | L3 Operational autopilot | Grade DFounder claims only | 1Observation date unknown · unclassifiedY Combinator | $193K (claimed)ARR · Observation date unknown · disputedFounder claim · disputed | €1.7M (parent Phospho pre-seed)As of 2024-01Tech.eu |
Evidence grades describe the sources behind reported facts, including financials. They do not certify autonomous operation. Levels are editorial assessments; human involvement still needs to be checked.
Data: The Autopilot Index (CC BY 4.0). Download: JSON API · OpenAPI · MCP. Full profiles and the watchlist are on /companies.
Cite this: The Autopilot Index, Autopilot Leverage Screener, autopilotindex.com/dealroom, accessed <date>.
Newsletter
Get Autopilot Pulse
Weekly signals from the businesses that run themselves: companies, field experiments and the tech stacks they run on. Every claim is labelled verified, self-reported or disputed. Free, and you can unsubscribe in one click.
FAQ
Questions
What is the Autopilot Leverage Screener?
A screen we're building at the Dealroom hackathon on Thursday 1 October 2026. It joins The Autopilot Index (companies that say AI runs them, each with an autonomy level and an evidence grade) to Dealroom's funding and headcount data, so funding per employee and revenue per employee can sit next to how much of the business AI actually runs. Until then, this page shows our own index data only.
What do L1 to L5 mean, and why is L1 left out?
They're the Index's published autonomy levels: L2 Function autopilot, L3 Operational autopilot, L4 Goal-level autopilot and L5 Full autonomy. L1 Copilot covers tools that help a human work faster. The Index only tracks businesses where AI performs the work, so L1 is out of scope. The one-line definitions are in the table on this page. Levels are editorial assessments, not audits.
Where do the numbers on this page come from?
Every figure on this page comes from The Autopilot Index: dated, source-linked observations that we record ourselves. Nothing on this page comes from Dealroom. If a figure has no source it isn't shown, and a blank cell means not disclosed. We never fill a gap with an estimate or a zero.
Can I use the data or query it from my own agent?
Yes. The Index data is CC BY 4.0: use it and credit it. There's a free read-only REST API (/api/v1, OpenAPI at /openapi.json) and an MCP server at autopilotindex.com/mcp. No login. Both serve our index data only.
What do I get if I sign up?
Autopilot Pulse: weekly signals from the businesses that run themselves, with every claim labelled as verified, self-reported or disputed. It's free and you can unsubscribe in one click.