Lead AI Product Manager

About Textlayer Trust is the central design problem of the AI era. Most companies treat reliability as something you bolt onto a demo. We treat it as the actual work: the part where promising prototypes meet real users, real data, and real consequences. We're a small remote team, building production-grade AI systems for PE-backed portfolio companies in compliance, financial services, FP&A, and logistics. Our methodology runs from Align → Build → Grow: scope what's worth building, ship it with evals and observability that hold up under load, and use production behaviour to find the next thing worth building. If you're excited to own client engagements end-to-end, build the artifact trail that makes AI systems reliable in production, and help define what eval-driven product work looks like, let's connect. The role The Lead AI Product Manager owns client engagements end-to-end: the discovery call, the eval rubric, the scoping doc, the production handoff, and the post-launch growth phase based on learning from real usage. Your job is to make sure the right AI system gets built. A year here compresses what most AI PMs get in five. Key responsibilities Lead Align: run discovery with operating partners to identify the use case worth building, define what "working" means in measurable terms, and write the go/no-go. Turn ambiguous AI ambitions into eval-ready scope. Turn "we want AI in our underwriting flow" into success criteria the engineering team can build against. Hold scope with senior stakeholders. When clients push for more features, faster timelines, or expanded surface area, you can push back productively while keeping the relationship strong. Work as a peer with senior engineers. You'll read their code, reason about RAG and agent architectures, evaluate eval rubrics, and tell a prompt problem from a system problem from a data problem. Run the engagement cadence: weekly client reviews, internal eval reviews, milestone demos, with artifacts that survive turnover on the client side. Drive post-launch growth: learn from what production data reveals, identify outlier behaviours, and help identify what's worth building next. Codify what works: the discovery script you write for one engagement becomes the team's default for the next. Mentor PMs as the team grows, and help raise the bar for what product work looks like at Textlayer. What you bring To succeed in this role, you'll need a strong PM foundation, real AI product shipping experience, comfort going deep technically with engineers, and the disposition to thrive in client-facing engagements. Required qualifications 7+ years of product management experience, with at least 2 years shipping AI products end-to-end. This means products where you owned the trade-offs between model behaviour, eval quality, latency, cost, and user need, and lived with the consequences when those trade-offs were wrong. Strong product generalist fundamentals before AI specialism. The harder parts of this job are the parts every great PM has always been good at: scoping ruthlessly, building trust with stakeholders who don't know what they don't know, and keeping engagements pointed at outcomes that matter. Comfort going deep technically. You don't need to ship code, but you can read a RAG pipeline, reason about retrieval failure modes, sanity-check an eval set, and call out when a "model problem" is really a data problem. Engineers should leave conversations with you sharper, not slower. Comfort with technical depth is non-negotiable; depth itself can be grown. Strength in client work. You're energized by sitting across from a CFO who doesn't fully trust AI yet and walking out with a scoped engagement and a clear win condition. You see the relationship as part of the product. Not all PMs are wired this way. If that energizes you, this role will keep giving; if it drains you, this isn't the right shop. Strong written communication. Most of the artifact trail of this role is written. The go/no-go memo at the end of Align decides whether an engagement happens, and it's yours to write. If writing well isn't already a habit, this role will be painful. Bonus points You've worked with PE-backed portcos or inside a portfolio company, and you know how an operating partner thinks. You've shipped in compliance, financial services, FP&A, or logistics, and felt the constraints these industries put on AI deployments. You've worked in a consultancy or services-led business and understand the difference between a billable engagement and a product roadmap. You've done real eval-driven development: open coding, error analysis, rubric design, and closing the loop from production behavior back into evals. You've built or torn down a methodology before. You know what survives contact with real engagements and what doesn't. You've owned commercial scope: writing SOWs, defending price, and expanding accounts. Don't meet these? Apply anyway. But tell us about something you shipped. Not just where it worked, but where it broke and what you learned. We care more about the second half of that sentence than the first. Strong applications from non-traditional backgrounds are welcome; weak applications from "the right CV" are not.