Make something genuinely useful. Solve a real problem. Attract users.
Subsidize growth. Get everyone on the platform. Become the default.
Take 30%. Sell the graph. Optimize for shareholders, not users.
Make it worse for everyone. Enshittification. You're locked in anyway.
Uber takes 30% and tells drivers what to charge. Airbnb inflates neighborhoods. LinkedIn sells your graph. The pattern is always the same.
The cost of compute is falling. LLMs can do what entire teams used to do. The scarce resource isn't intelligence anymore.
Who do you let make decisions for you? Where does that computation happen? Who sees the data? These are the questions that matter now.
The innovation isn't the problem. The foundation is. Every generation throws more money and compute at the same extractive model — cut flowers that look beautiful but are dying from the start. The consumer landscape is evergreen. You just have to plant in actual soil.
In orbital mechanics, a gravity assist uses the momentum of a nearby body to change your trajectory without burning fuel.
Every meaningful decision you make works the same way. Your doctor. The neighbor who knows the good plumber. The friend who just got back from Portugal. Their knowledge changes your trajectory.
AI can scale this — but only if the trust relationships are real, not manufactured by a platform selling your attention.
Allow me to notice my friends' momentum.
Follow them. Slip into their wake.
Not because an algorithm decided I should — because I trust where they're going.
This is the core primitive. Everything else is an application of it.
People don't think in words. They think in proximity, trajectory, weight. Concepts cluster. Decisions trace paths through possibility space. Values live in a kind of mental topology.
Language is a lossy compression of how people actually reason. It's why surveys lie, why focus groups mislead, why people can't articulate what they want but know it when they see it.
Threshold's core technical bet: knowledge graphs and manifolds that capture how people actually think — not what they say they think. Spatial reasoning that transcends language and gets closer to values.
Trust relationships as weighted edges. Domain-specific. The same person might be a 0.9 for restaurant recs and a 0.2 for investment advice. Context shapes the topology.
Decision space isn't flat. It curves around values, constraints, experiences. We're trying to reverse-engineer people's mental models — not their stated preferences.
The protocol is open. Best-in-class first-party offerings for what you trust us with. Total control over what you don't.
Simple transactions with mutual benefit in low-trust environments. Prove you qualify without revealing why.
Run inference on your hardware. Your data never leaves your machine unless you say so.
Services + equity. Build alongside partners — consulting engagements paired with equity in projects that extend the protocol.
Every engagement validates a vertical. AI discoverability, audience simulation, organizational analysis — each one stress-tests the stack.
Revenue from day one. IP compounds with every project.
Audience management tools that make it rational for consumer app builders to share audiences on the platform.
Every app that joins makes the network more valuable. Developers get distribution. Users get portability. The protocol captures the coordination premium.
Platform fees on coordination, not extraction. Subscriptions, not ad revenue.
Federated coordination at conglomerate scale. Cross-subsidiary data sharing with trust boundaries intact.
Loyalty programs spanning telecom, commerce, banking, and travel. Unified customer intelligence without centralized data lakes. Each division keeps sovereignty.
Enterprise licensing. The coordination layer for organizations too complex for monolithic platforms.
Data sovereignty is the consumer guarantee. The platform captures value through coordination, not extraction.
The Audit scores how AI models see a business — across providers, with and without web search. The Simulator generates projected audience reactions to actual content.
Personalized learning paths that adapt to how a student actually thinks — not demographic buckets. Trust-weighted recommendations from mentors and peers.
Evaluate candidates the way organizations actually make decisions — through simulated team dynamics, not keyword matching on resumes.
Simulate how information flows through an org. Find the bottlenecks, the silos, the invisible decision-makers. Test restructuring before you do it.
Agents that operate within trust boundaries you define. Not autonomous in the "hope it works" sense — autonomous in the "I set the rails" sense.
Your browsing history, your bookmarks, your reading patterns — analyzed locally, for you. Surface what you actually care about, not what an algorithm thinks you should.
Every one of these showed up in a single conversation. They're all the same pattern.
Driver sets the price. Pays a flat subscription, not 30% per ride. Open platform. Rider finds drivers through trust networks, not a monopoly app.
2–3 degrees of separation. Your friends' lake houses, beach houses, mountain places. Coordinate calendars. Equitable exchange. No Airbnb tax.
When someone goes somewhere, they create recommendations for their network. Networks grow organically from in-person conversations. People share what they have.
Aggregate wilderness & camping permits for the state of Colorado. Select areas you like. Pool with friends. Find and book across fragmented systems.
Allow me to notice where my friends are going. Slip into their wake. Not surveillance — shared intention. Opt-in gravity assists.
Expose the car as a node in the network. Parking spaces in Vail. Cleaning services as indexers. Physical assets become part of the coordination layer.
People who trust each other — real relationships, not platform connections.
Data they'd share with those people — but not with a platform that sells it.
Physical and digital worlds intersecting — cars, houses, permits, parking, calendars.
Coordination that benefits everyone — not just the middleman.
A VC-funded platform captures the coordination.
Takes a cut. Sells the data. Optimizes for growth.
Eventually makes it worse for everyone.
You can't leave because your network is trapped there.
The coordination runs on open rails.
Your data stays yours. Your graph is portable.
Applications compete on experience, not lock-in.
Trust is the protocol. Everything else is an app.
Let platforms commoditize your intelligence, your relationships, your attention. Accept the 30% tax. Hope the algorithm is kind.
Own the rails. Choose where compute happens. Keep your graph. Let trust — not lock-in — determine who coordinates your world.