Senior Product Manager, Agentic Platform

The Role Fetch is seeking a Senior Product Manager for our Agentic Platform to define and lead the internal platform that Fetch's AI agents are built on — the shared runtime, context, evaluation, and governance foundation that turns agent ideas into production systems. This is a highly strategic and technical role at the intersection of emerging AI capabilities, platform engineering, and the teams across Fetch who are building agents into their products and workflows every day. You will own two complementary mandates: (1) the Agentic Platform portfolio — the shared capabilities every agent team at Fetch builds on, spanning agent runtime and orchestration, context and memory, model routing, evaluation, observability, and governance, plus the agent development lifecycle (ADLC) that takes an agent from idea to production — and (2) internal user experience and adoption — deeply understanding the engineers, product managers, and data teams who will depend on the platform to build, ship, and operate agents, and ensuring what we ship genuinely works for them. That second mandate is unusual and deliberate: all of your most important users are your colleagues. You will regularly flex into a UX-researcher mode — sitting with agent teams as they build, shadowing an agent's path from prototype to production, and mapping the development lifecycle end to end — so that the platform we build becomes load-bearing in real daily work rather than an impressive demo. This role requires strong product judgment, genuine technical fluency in modern AI (LLMs, evaluation frameworks, agentic loops), exceptional organization, and the ability to navigate ambiguity while helping define how an entire company builds with agents. You will partner cross-functionally with Engineering, Data, Design, Security, and the agent teams across Fetch to ship platform capabilities that drive measurable impact for the teams building on them — and for the business their agents serve. You'll be joining Fetch at a pivotal moment: FAST, our publicly launched AI insights platform, has already put agentic tools in the hands of advertisers and our sales organization, and teams across the company are building real automation muscle. We're moving from ad hoc agent experiments to an intentional, shared agentic foundation — and this role owns that foundation as a product. WHAT YOU'LL DO Agentic platform products * Own the product vision, strategy, and roadmap for Fetch's internal Agentic Platform: agent runtime and orchestration, context and memory, model routing and gateways, evaluation, observability, and governance. * Treat the platform as a product: prioritize capabilities by what agent teams actually need next, sequence the roadmap against real agent launches, and measure success by what ships on top of the platform. * Design the platform's human-in-the-loop and multi-agent primitives — the shared building blocks teams use to define roles, handoffs, review points, escalation paths, and feedback loops between agents and the people who oversee them. * Partner with Fetch's developer productivity and infrastructure teams so the platform shows up to builders as one coherent experience — shared capabilities and paved paths rather than one-off solutions. * Extend what works: as capabilities prove out with one agent team, generalize them into shared primitives other teams can adopt — building on the automation muscle Fetch has already developed. * Define success metrics tied to platform outcomes — time from agent idea to production, adoption across teams, reliability, and quality and cost per agent task — and drive experimentation to improve them. * Look around the corner: anticipate where the agent ecosystem is heading — MCP, agent-to-agent interop, agent identity and delegated access — and position the platform so Fetch's agents can safely work with agents and surfaces we don't control. Agent product development lifecycle * Own the ADLC end to end: define how a team at Fetch takes an agent from idea through evaluation, launch, and production operations — and make that path faster, safer, and more repeatable with every release. * Build and operationalize evaluation capabilities every agent team can use — offline evals and golden datasets before launch, production quality monitoring after. * Define the shared building blocks of the lifecycle: prompt, schema, and tool registries; a unified programmatic interface to the platform (CLI, API, MCP); and onboarding checks for new agents and tools. * Design governance that enables rather than blocks: least-privilege agent access, guardrails, and review points that let teams ship quickly and trust what they ship. * Drive continuous improvement loops: prompt and context engineering, model updates, and tool design informed by eval results and real user feedback. * Establish observability and performance metrics for agent effectiveness, reliability, and cost across the portfolio. Internal discovery, research, and adoption * Serve as the embedded researcher and voice of the teams who build on the platform: engineers, product managers, data scientists, and the operations teams who depend on what they ship. * Run structured discovery — interviews, ride-alongs, workflow shadowing — and translate it into journey maps of the agent development lifecycle, friction inventories, and prioritized platform opportunities. * Decide where the platform should be opinionated and where it should stay out of the way — defining the defaults, escape hatches, and override mechanisms builders need to trust the platform with production workloads. * Own adoption as a first-class product outcome: onboarding, enablement, feedback channels, and iteration until the platform is what teams reach for by default — not just tried once. * Close the loop between builders and the roadmap, ensuring friction, escalations, and workarounds feed directly into platform improvements and evaluation sets. Across the portfolio * Run the program with rigor: crisp documents, roadmaps, and status communication; dependencies tracked across many teams; nothing dropped. * Build alignment across the many stakeholders this work touches — engineering, data, security, design, and the agent teams building on the platform. * Translate complex technical concepts into clear business value for leadership — and translate builder and operator realities back into technical requirements for engineering. * Communicate progress, risks, and tradeoffs clearly to senior leadership. * Operate with a high degree of autonomy in a space where the problems, tools, and org structures are all still taking shape. MINIMUM QUALIFICATIONS * 4+ years of Product Management experience, with meaningful time spent on platform or infrastructure products, developer tools, or AI/ML-powered products. * Experience shipping AI/LLM-powered features to real users, with working fluency in LLMs, prompt engineering, evaluation frameworks, and agentic loops. * Demonstrated discovery and research skills — comfortable running your own user interviews, shadowing sessions, and workflow mapping, and turning them into product requirements. * Strong systems thinking and process design capabilities, with a track record of identifying and redesigning complex workflows. * Exceptional organizational skills: able to run multi-team programs, manage dependencies, and keep many stakeholder groups aligned simultaneously. * Strong analytical mindset with experience defining metrics, running experiments, and driving measurable outcomes. * Proven ability to work cross-functionally and influence stakeholders across technical and non-technical teams — especially engineering and data science. * Excellent written and verbal communication skills, with the ability to clearly articulate complex ideas to diverse audiences. * Comfort operating in fast-paced, ambiguous environments where problem spaces are not yet fully defined. PREFERRED QUALIFICATIONS * Experience building internal platforms or developer-facing products where your users are other builders — including measurable adoption, enablement, and change management. * Experience working with AI agents, multi-agent systems, or conversational AI in production environments. * Familiarity with the emerging agent infrastructure ecosystem: agent frameworks, MCP and agent-to-agent protocols, model gateways, and agent observability tooling. * Familiarity with AI evaluation frameworks, prompt engineering, and quality measurement for generative AI products. * Experience designing human-in-the-loop systems where AI output is reviewed, corrected, or approved by expert operators. * Formal or informal UX research experience: study design, contextual inquiry, usability testing, or workflow analysis. * Background in marketplace, rewards, loyalty, or retail media businesses.