Founding Product Lead

• Three knowledge graphs one answer. A Commerce Graph (what exists prices availability) a Truth Graph (product claims that survive our verification research and carry citations) and a Preference Graph (what each user actually cares about). The Preference Graph is the least-built of the three - you will likely own its birth. Every product decision is a decision about how the three combine into one trustworthy answer.
• A chat product whose edge is structural. General assistants average the internet confidently. Ours answers with verified truth and citations - including "don't buy this." The bar isn't engagement; it's users trusting the verdict on a high-stakes purchase. Calibrated trust is a harder problem than delight.
• Agent-era distribution. The product has to be a named capability inside ChatGPT Apple App Intents Gemini and Claude - surfaces where an AI agent not a human decides whether to call us. Being chosen by agents is the new being indexed by Google and you'll write that playbook while you ship it.
• The build pipeline. Intent becomes a visual mockup then a locked spec then agents build then separate verifier agents check the build against the spec. The verifier is the load-bearing part: it's a test the building agent can't author or grade so the system can't quietly approve its own work. Your spec is what those checks run against. Your taste gates what ships.
• Cortex the company that compounds. You'd work inside Cortex - the shared AI brain that runs the company and is the same product family we sell. Every operator works through governed AI sessions and our agents answer their own questions from a shared substrate of 8600+ documents. Nobody else builds this way. You need to model where the product is heading and write specs that anticipate direction - reading the system ahead of the brief rather than waiting for one.

• Product law. The locked specs agents build from across chat web extension mobile and personalization. Agents write the code and content; you write the spec the verification and the verdict on what ships.
• The consumer quality bar. The standard for everything a shopper touches held as evidence tests for four to six product outcomes - each falsifiable each with a test a stranger could run. When the bar and the schedule conflict you hold the bar.
• The decision-shaped UX standard. With our Founding Designer you'll define what a verdict looks like: interfaces shaped around the decision the user is making - buy wait walk away - rather than around the content we happen to have. Communicating calibrated confidence (when we're sure when we're not) is the core interaction problem.
• Outcomes personally. One or two falsifiable product outcomes a quarter that you run end-to-end yourself. Visibility here is decisions registered and outcomes moved - the work speaks for itself not the hours behind it.
• The operator contract. This is the model we run: in your first quarter you co-sign a seat charter - one machine-checkable number that proves the seat works plus a written split of what you decide freely and what you bring to the founder. The seat is defined on paper then you own it.

• You independently form working models of complex systems notice where your model is wrong and update fast - including killing your own ideas when the evidence says so. You don't need perfectly defined scope to start; you need enough signal to reason from first principles. You write clearly because clear writing is evidence of clear thought and here your writing is executable.
• You move between strategy and locked spec without getting stuck at either altitude - from "what should chat do when the evidence conflicts?" to a spec agents can build from same day. Agents are your production system and you verify what comes back: you can do this job by hand and prove it and that mastery is exactly what lets you trust - or reject - the verdict an agent hands you. You treat the agent's output as something you check not something you accept on faith. The expensive thing here is a redo cycle never the compute.
• You've shipped consumer products people actually use - live surfaces not strategy decks - and written specs precise enough that someone or something built the right thing without a meeting. Commerce marketplaces search or conversational products are familiar ground. We care about the artifact and the reasoning more than where you did it.
• Who this isn't for. This role is wrong if your product practice is roadmap theater - decks alignment meetings and planning rituals that never touch the build. It's wrong if you need an engineering team to hand you velocity or a PM org beneath you to feel senior. It's wrong if what you're chasing is a VP title and the team-building that usually comes with it - here you direct agents and a handful of elite operators and your judgment in writing is the thing that ships. It's wrong if you're comfortable shipping what an agent produced without being able to say why it's right. You'll be happiest here if you want to own the whole product loop yourself and be measured on what it produces.

What We Do

Product.ai (formerly Demand.io) is the truth layer for commerce. Built on Axiomatic Intelligence — a proprietary adversarial reasoning methodology that stress-tests product claims against physics economics and engineering constraints — Product.ai delivers verified purchase verdicts not summaries. Product.ai tells consumers when NOT to buy. Product.ai emerges from Demand.io a profitable bootstrapped AI commerce company whose SimplyCodes platform processes over $1B in annual transaction value with a team of 20. Founded by Michael Quoc.

Gallery