AI Agent Product Manager

Hello Patient is transforming healthcare communication through advanced conversational AI. They are seeking a technical, high-agency AI Agent Product Manager to own the end-to-end delivery of AI agents in real customer environments, managing deployments, customer requirements, and product launches.

Responsibilities:

• Lead delivery of net-new AI agent deployments as the single accountable owner—from customer requirements through production go-live
• Treat every deployment as a product launch: define scope, manage tradeoffs, hit timelines, ensure quality
• Own the customer-facing roadmap for each deployment—what are we shipping, in what order, and why
• Gather requirements directly from customers and translate them into agent behavior yourself using our internal platform—prompts, tools, guardrails, multi-agent workflows
• Do hands-on prompt engineering using AI tools to build the components that make agents work: conversation flows, decision logic, edge case handling, integration behaviors
• Ship production-ready agents that handle real healthcare complexity: scheduling logic, patient context, EMR data, practice-specific edge cases
• Continuously improve agent performance using structured evals, live call review, and direct customer feedback
• Lead deployments for new workflows, integrations, and verticals where no playbook exists
• Identify unknowns early, derisk them, and make tradeoff decisions that keep launches on track
• When deployments require net-new platform capabilities, own requirements and prioritization with the engineering team
• Conversational designers design and optimize agent behavior—they review real conversations, identify patterns, design conversation flows, and ship agent behavior
• Customer engineers build integrations into our customers' systems of record—EHRs, PMS, and PIMS
• You own everything else: integration scoping with Customer Engineering, cross-team coordination, timeline management, stakeholder communication, and unblocking whatever's in the way
• Identify patterns across launches and codify them into playbooks
• Know when a workflow or integration is ready to transition from zero-to-one ownership to steady-state implementation
• Hand off proven patterns to implementation teams and move on to the next net-new challenge