AI fact layer that turns portcos into autonomous businesses.
Hand more of each portco's work to agents that run on approved facts, not hallucinations. Generate a hypothesis tree from those facts and steer every company to success.
Funds are becoming product studios.
AI made software cheap to build. Funds can now build software companies themselves, not just back them.
Capital is concentrating in hardware and deep tech.
Cheap software is easy to copy, so the moat moves to hardware and deep tech. Capital is following it.
Autonomous portcos need institutional design.
Agents are replaceable. The rules agents work under shape outcomes more than the model or harness. Get them right and the mission holds.
Funds use AI to pick companies. They should use it to run them.
Sourcing and diligence are automated. Post-investment is still decks, email updates and quarterly calls.
Lunar VenturesFund partner
Fund partner
PartnerSteering moves from opinion to evidence.
An agent reads each company's event log and infers its stated and revealed bets. Every node carries its evidence and status, so you can model the business against market data and benchmarks.
- Change logs
- Press releases
- News events
- Investor updates
- Live API sources
- Fact layer (highest resolution)
We believe builders are the primary customer segment.
We believe agencies convert faster than solo builders.
We believe the team can ship major rebuilds quickly.
We believe agent features ship weekly without regressions.
We believe usage pricing plus resellers grows revenue.
We believe builders are the primary customer segment.
We believe agencies convert faster than solo builders.
We believe the team can ship major rebuilds quickly.
We believe agent features ship weekly without regressions.
We believe usage pricing plus resellers grows revenue.
Growth, product, ops and leadership use agents via MCP on the same facts.
A fact is something true about the company: approved, versioned and served from one API. Not memory an AI picked up along the way that nobody signed off.
Deploy agents across every function and connect them over MCP. Each one works from the same source of truth.
Campaigns that quote current pricing
Specs written against the live roadmap
Vendors and runbooks from one record
Board updates from approved numbers
Campaigns that quote current pricing
Specs written against the live roadmap
Vendors and runbooks from one record
Board updates from approved numbers
The bets, made visible
Generated from the company's facts and tested against live market data and comparable rounds. Works standalone.
One source of truth per company
Approved, versioned facts served from one API. Every team and agent connects over MCP. Built on NOAN.
Mission-lock through later rounds
The mission written into the company's structure, with non-economic entities for oversight, so it holds as new capital and board seats arrive.
Agents bound by organisation code as law
Company-specific rules enforced by the environment, not written into prompts. Agents comply because they are checked.
Portfolio steering
Trees across the portfolio, built from events you already receive. No founder onboarding.
Company fact layer
The full ontology inside a company, with MCP access for every team. Start with one.
Build a hypothesis tree for a portco today.
Add what you already have. Generate the first tree from it.