You own the whole product line at a profitable company rebuilding shopping around verified answers: the bar, the specs, and the call on what ships. Product.ai is the verified truth layer for shopping, the intelligence that tells you what's actually true about a product, including when not to buy. Our first proof at scale is SimplyCodes, the code verification service, at roughly $22M in revenue. Profitable. Bootstrapped since 2009. No outside investors. No board. Fewer than twenty operators outbuilding companies 10× our size. Why This Role ExistsYou will author more of the shipped product than anyone except the founder. Our founder owns product strategy, and that doesn't change. What's open is the seat below it, and the question it exists to answer is which Product.ai surface earns the company's next dollar. Maybe conversational answers, the Truth Graph pages that carry them into search, personalization, or the paid verified-data capability developers and agents already call. You gather the evidence and make the case, then own the answer end to end. One constraint is settled; you inherit it rather than relitigate it. Truth Graph pages are SEO-first. A verified answer wins by being found, the same way SimplyCodes wins today. A spec here is not a document that drifts out of date. A locked spec is the roadmap-of-record. Agents build directly from it. Write the spec, and the product follows. This is a founding individual-contributor role, not a VP role with a PM org to build. You direct agents and work beside a handful of operators who each run an entire domain. You have to be the best spec author in the company. In a system where anyone who can write a clear spec can ship, the work flows to whoever writes the clearest one. It's a meritocracy of the written artifact, and you want it that way. Your leverage is judgment and taste instead of headcount. What to build, and how you'll know it worked. The System You'll Need to Model Knowledge-graph products. One trustworthy answer has to combine what exists and what it costs, which claims survive verification with citations intact, and what this particular shopper cares about. Every product decision is a decision about how those layers combine. Trust-calibrated conversational products. General assistants average the internet confidently. Ours answers with verified truth and citations, including "don't buy this." The bar is whether users trust the verdict on a high-stakes purchase, and engagement metrics don't measure that. Calibrated trust is a harder product problem than delight. Agent-mediated distribution. The product has to be a named capability inside ChatGPT, Apple App Intents, Gemini, and Claude. On those surfaces an AI agent rather than 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. We already sell verified commerce data to developers and agents on paid keys, so the surface question includes an API's product design as well as a consumer screen. The commercial relationship with those developers and platforms belongs to our Head of Commercial; the product shape of the capability they call belongs to you. Spec-driven agentic development. 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 crux. It's a test the building agent can't author or grade, so the system can't approve its own work. Your spec is the interface the whole pipeline builds and grades against. Your taste gates what ships. Cortex, where company knowledge builds on itself. You'd work inside the shared AI brain that runs the company, the same product family we sell. Every operator works through governed AI sessions, and our agents answer their own questions from a substrate of thousands of 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 instead of waiting for one. If reading that energizes you, keep going. If it feels overwhelming or underspecified, this isn't the right fit. What You Will OwnProduct law. The locked specs agents build from, across conversational answers, the Truth Graph pages that carry them into search, and personalization. Agents write the code and content; you write the spec, the acceptance tests, and the verdict on what ships. This is the discipline of treating the spec as the build interface for AI agents. Here it is the daily production system, well past the pilot stage. The consumer quality bar. The standard for everything a shopper touches, written as evidence tests for four to six product outcomes. Each is falsifiable, with a test a stranger could run. Quality-bar ownership and evaluation design are the transferable craft; our surfaces are the instance. 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, meaning 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, and it is decision-support UX at its hardest. Outcomes, personally. One or two falsifiable product outcomes a quarter that you run end-to-end yourself, from intent to locked spec to verified build. Visibility here is decisions registered and outcomes moved. The work speaks for itself. 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.
Who You Are:
You form your own working models of complex systems and notice where they're wrong, then update fast, killing your own ideas when the evidence says so. You don't need perfectly defined scope to start, just 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 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, and compute is cheap. You've shipped consumer products people use, live surfaces instead of strategy decks, and written specs precise enough that someone, or something, built the right thing without a meeting. The craft you must already own is spec authorship, consumer product judgment, and evaluation design. We accept commerce, marketplaces, search, or conversational products as comparable experience. Here you'll grow into directing agent fleets as a production system, verification design, knowledge-graph product architecture, and distribution through agent platforms. 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, meaning 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 comes with it; here, your judgment in writing is the thing that ships. It's wrong if this is an 18-month stop that writes a resume bullet. The seat charter is a multi-year bet in both directions. 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.
How We Evaluate:
Video screen. Brief and async — about 15 minutes, whenever works for you. It replaces the recruiter screen; answers go straight to the hiring team. We want to see how you think, not how you present.
Calls with company stakeholders. Short conversations with key members of the team.
Conversation with the founder. How you model systems, where you push back, and whether you can hold the product bar in a live argument.
Paid work trial. A paid 10-14 day strategic trial — real work in our real environment, taking a live product problem from intent to locked spec to verified build. We watch four things: how you get grounded, whether you write the spec before the build, how you verify what your agents produce, and whether your self-assessment is honest.
If the work above reads like yours but your resume is unconventional, apply anyway. We hire on the work and the reasoning, not the pedigree.
Compensation & Ownership:
Total first-year comp: $325,000 – $475,000(base + performance-based ownership and profit-share programs).Base: $250,000 – $300,000 .Beyond base: eligibility for the company's ownership and profit-share programs — grants are performance-based, terms discussed at the offer stage; 100% family premium coverage; and an effectively unlimited token budget, steered by ROI, never capped. This is a partnership structure built to mint partners. When the company wins, you win — in real, liquid dollars, every year. Based in Santa Monica, Los Angeles — in person, five days a week. The rooms are real rooms.
