The match graph for AI. The longer it runs, the stronger it gets.
Discovery is the bottleneck of this era. Thousands of AI artifacts exist — products, agents, MCPs, models, repos, workflows — but finding the right one for your specific need is broken. The web indexes pages, not capabilities. It ranks incumbency, not quality. We exist to change this.
Fragile
Damocles
A sword overhead. Any shock shatters the illusion. Static directories, listicles, and rankings that degrade the moment the landscape shifts.
Robust
Phoenix
Rises from ashes but only returns to baseline. Endurance without growth. Necessary, but not sufficient.
Antifragile
Hydra
Cut one head, two grow back. Disorder is not survived — it is metabolized into strength. This is what we build.
The Thesis
Cursor isn't one entry — it's fifteen capabilities. Every AI artifact is decomposed into what it can actually do, for whom, under what conditions. Intent matches to capabilities, not to brand names or marketing pages.
Every search produces a match record: intent, context, capability, artifact, outcome. Every failed match reveals a gap. After a year, the graph knows more about the AI capability landscape than any human, company, or LLM. The Lindy effect applied to intelligence.
Most systems optimize for the expected case and break on the unexpected. We treat the unexpected as the primary input for growth. Failed matches, edge cases, novel intents — the raw material of antifragility. Gap records become demand signals for builders. Disorder feeds the graph.
Smarter with every search.
Search the graphCommon Questions
The match graph for AI. We map every AI artifact in the world — products, agents, MCP servers, APIs, repos, models, extensions, workflows — decompose each into discrete capabilities, and match them to human intent. Every interaction makes the graph smarter. Rooted in Nassim Nicholas Taleb's antifragile framework: systems that grow stronger through disorder.
Artificial intelligence that improves from stress, volatility, and disorder. Based on Taleb's framework: fragile breaks under stress, robust survives stress, antifragile gains from stress. Our match graph is antifragile — every failed match reveals a gap, every search refines confidence scores, every gap record becomes a demand signal for builders.
Directories list products. We decompose them into capabilities and match those capabilities to your specific intent and context. When you search, you get one confident answer — the single best artifact for your situation — not ten links to sort through. Rankings are based on real match signals, not who pays the most. Sponsored results are always labeled and never influence organic matching.
A knowledge graph where nodes are capabilities (what AI artifacts can do), edges are match records (intent + context → capability → outcome), and intelligence emerges over time as the graph learns which capabilities solve which intents for which user profiles. It compounds. After a year, the graph knows the AI capability landscape better than any human.
A composite of match confidence, editorial quality, freshness, click-through rate, builder verification, and relevance to your specific intent. No pay-for-position. Sponsored placements are separate from organic rankings and always clearly marked.