Head of Product, Agents Job at Oscer in San Francisco

Full time The AI startup growing faster than Canva That's what the Financial Review called us. Two years ago, we pivoted from broad healthcare AI to building Earths finest clinical exoskeleton. Our mission is simple: double care capacity by eradicating non-clinically-useful busywork. We've found product-market fit with individual clinicians through our freemium medical scribe, transforming unstructured clinical visits into structured text artefacts. Every new job a clinician delegates to Heidi makes patients feel more attended to, cleans up health system logjams, and lets clinicians be clinicians again. We're looking for a Product Manager who has built and shipped agent-based AI systems. You will own the evolution of our AI capabilities from reactive assistance to proactive automation, defining how intelligent systems can safely and effectively operate in clinical environments. This role will be based in our San Francisco office. We don't care about logos; Architect agent systems that can reason through multi-step problems, understanding context and making intelligent decisions about tool use Build evaluation methodologies for agent performance, measuring accuracy, reliability, and real-world utility in production Establish safety and constraint systems that ensure AI agents operate within appropriate boundaries Create feedback loops between agent behavior and product iteration, using real-world performance to drive improvements Shipped an experiment that challenged a core assumption about clinician behaviour Jumped into customer support tickets to understand current limitations 4+ years PM experience with at least 2 years building AI/ML products, specifically agent systems, autonomous workflows, or multi-turn AI applications - but we care more about what you've built than years on the clock ~ Deep technical expertise in agent architectures - you've worked with frameworks like LangChain, implemented reasoning patterns like ReAct or Chain-of-Thought, and understand tool use in production ~ A romantic streak about software - you believe products should be beautiful, not just functional, and that great design can transform someone's day ~ Ability to design evaluation methodologies for agent systems, including performance monitoring and failure analysis ~ The disagreeableness to say the thing that shifts the entire room's perspective (and the tact to say it well) ~ Data fluency - you run your own queries, generate your own insights, and make decisions based on evidence ~ Track record of shipping products with measurable impact on real humans ~ Comfort with ambiguity and ability to navigate regulatory and compliance requirements in sensitive domains Can you debug why an agent made a specific decision by analyzing its reasoning trace? Are you able to execute without a legion of data analysts, product marketers, and research coordinators at your beck and call? Systems are log-jammed, clinicians' lives degrade in quality, and patients churn through the meat-grinder. The bad news is its not an easy job - you will tear your hair out more than once. The good news is you will trust and admire everybody you work with, and rest easy knowing youre doing THE defining work of your career. We offer additional paid days off for your birthday and wellness, corporate rates at Anytime Fitness, a $500 annual development budget, and serious equity. We are flexible, and value our teams mental health & overall wellbeing. The nature of the scale-up game is demanding, but can adapt to your schedule. We operate on a trust-only policy: company OKRs individual KPIs.