Patronus AI is seeking a Senior Product Manager to own the roadmap for the internal platform, the environment platform, and the tooling that enable a small team to deliver RL environments. The role covers the hosted environment platform, agentic tooling for building and QA-ing application clones, expert onboarding, delivery tracking, and the spec pipeline that feeds engineering. The candidate will design agent-centric workflows where agents are primary users and humans are the exception. You will write versioned, frozen PRDs at kickoff with MVP scope and explicit cut lists, and ground decisions in real usage rather than opinion. You will work with internal customers—environment engineers, SME operations, QA, and GTM—to translate learnings into a ranked backlog and clearly explain what will not be built and why. You will use usage data to inform decisions, instrument tools, and retire features that do not earn their keep. You will close the loop between QA and spec, fixing pipelines when missing features are surfaced. You will apply the agent-first principle to your own work, building agentic workflows for spec generation, review, and versioning, and focusing your human time on judgment calls only you can make. You will stay hands-on, running agents in the environments, prototyping with AI, and building small evaluations to verify feature effectiveness. Stay ruthlessly scoped: ship the smallest thing that changes an internal customer’s week, publish the cut list, and iterate from evidence. This role is based in our San Francisco headquarters and requires in-office attendance

Patronus AI is seeking a Senior Product Manager to own the roadmap for the internal platform, the environment platform, and the tooling that enable a small team to deliver RL environments. The role covers the hosted environment platform, agentic tooling for building and QA-ing application clones, expert onboarding, delivery tracking, and the spec pipeline that feeds engineering. The candidate will design agent-centric workflows where agents are primary users and humans are the exception. You will write versioned, frozen PRDs at kickoff with MVP scope and explicit cut lists, and ground decisions in real usage rather than opinion. You will work with internal customers—environment engineers, SME operations, QA, and GTM—to translate learnings into a ranked backlog and clearly explain what will not be built and why. You will use usage data to inform decisions, instrument tools, and retire features that do not earn their keep. You will close the loop between QA and spec, fixing pipelines when missing features are surfaced. You will apply the agent-first principle to your own work, building agentic workflows for spec generation, review, and versioning, and focusing your human time on judgment calls only you can make. You will stay hands-on, running agents in the environments, prototyping with AI, and building small evaluations to verify feature effectiveness. Stay ruthlessly scoped: ship the smallest thing that changes an internal customer’s week, publish the cut list, and iterate from evidence. This role is based in our San Francisco headquarters and requires in-office attendance