Definition
The matching problem
Selecting the best small set from a huge inventory using many inputs at once.
From the episode “The Self-Driving Car of Men's Fashion”Hear it on the podcast1 episode
Learn the language of AI-native founders
Every week: the new terms, tools, and workflows founders are using, straight from the podcast.
Related terms in Product & Engineering
The 80–90% Wall (Vibe Coding)The point in a vibe coding session where forward progress stalls on a specific unresolvable blocker. The UI is working, the demo looks good, but one thing — a database query, an integration, a specific feature — won't respond to any prompt variation. You've tried ChatGPT, Perplexity, YouTube tutorials, re-prompting 47 different ways. Energy leaves. Project dies. The diagnostic: if you've tried the same thing three or four different ways and you're still stuck, you've hit the wall and need a real human with domain-specific tool knowledge.The Buzzing Phone LoopA product design principle for non-desk workers: route every meaningful product event — new lead, new review, new booking — to a phone push notification rather than an email dashboard. For contractors who never check email and may not own a laptop, the phone buzz is the product's entire UX. It creates daily engagement and tangible perceived value from a tool the user might otherwise forget is running in the background.The End of Enterprise UIThe thesis that enterprise business application interfaces — forms, dashboards, approval workflows, multi-step navigation designed for human clicks — will be progressively replaced by database-plus-chat architectures. The interface becomes voice, text, or messaging; users describe what they need and the system retrieves, processes, and presents. Ran's timeline: 2–3 years for most business process applications. Leisure UI (browsing, shopping, social) survives because humans want the experience. Business process UI winds down because the cognitive overhead of navigating it is work AI handles better.The Last Mile (App Building)The final 10–20% of app completeness that current AI tools cannot reliably cross alone: custom integrations with third-party services, edge-case business logic requiring judgment, and complex UI/UX flows that don't follow standard patterns. The last mile is why production-quality AI builders offer human assistance plans: AI generates a structurally sound, largely functional app; humans close the gap between 'mostly works' and 'production-ready.' Analogous to self-driving cars' difficulty with edge cases — the 80% is solved, the 20% requires the hardest work.Tier 1 SupplierProvides a critical component or subsystem integrated into another company's product.Tier One SupplierA component manufacturer that supplies directly to a system integrator or OEM, rather than building the end product. Seven Sense positioned itself as a tier one supplier of sensing and compute to robot manufacturers — selling into the entire market rather than betting on one product category.Time to InsightThe elapsed time between a business question being asked and a reliable, actionable answer being available. AI analysts reduce this from days to under 90 seconds.TolerancesThe permitted variation in a part's dimensions. Tight tolerances mean parts fit consistently across thousands of units.