Lead AI Voice Agent Product Manager

Hello Patient; United States; On-site; USD 110,000 - 145,000; Full time

Job summary

Hello Patient is hiring an AI Agent Product Manager with a strong background in building AI voice agents from scratch. You’ll design how our agents listen, interpret, respond, and guide patients through real healthcare workflows, turning complex requirements into conversations that feel natural and accomplish the task.

This hands-on role requires creating prompts, conversation flows, and agent configurations yourself, then using simulations and real chats to identify and improve issues.

Qualifications

• 1+ years in conversational design or voice UX with hands-on AI agent building.
• Experience designing, testing, and refining how agents behave in real conversations.
• Strong understanding of conversation structure, turn-taking, intent recognition, ambiguity, clarification, and recovery when an interaction goes off track.
• Hands-on prompt engineering translating customer workflows into agent instructions, conversation flows, and decision logic.
• Ability to review calls and transcripts, identify patterns in conversational failures, and turn those findings into improvements.
• Technical fluency with agent-building platforms, tool calls, APIs and integrations, including how system constraints affect the conversation.
• Strong ownership and ability to work across customer-facing and engineering teams in a fast-moving environment.

Responsibilities

• Design and build production-ready agents.
• Own agent architecture and configuration, including prompts, intents, conversation flows, tool calls, extensions, transfer protocols, and retrieval-augmented knowledge bases.
• Lead technical discovery to translate customer business processes, rules, exceptions, and handoffs into agent behavior.
• Build agents that handle real healthcare complexity, including scheduling logic, patient context, EHR data, and practice-specific requirements.
• Partner with Engineering on integrations, new features, and end-to-end testing, translating integration constraints into agent design decisions.
• Assess technical feasibility and help the Customer Delivery PM understand implementation effort and tradeoffs before scope and timelines are committed.
• Own agent quality and production performance.
• Author evaluations, tags, and simulation scenarios that test expected behavior, edge cases, and failure modes.
• Build and run simulation suites, review results, and own internal agent quality before customer demos and testing.
• Peer-review other builders' agents and take accountability for the technical approvals you provide.
• Diagnose issues during soft launch and the initial period after go-live, using live conversations, evaluations, and customer feedback to guide improvements.
• Decide when an issue requires a targeted fix and when the agent needs to be re-architected.
• Guide agents as they evolve.
• Author the QA Agent Playbook and prepare Agent Technician and QA teams to support each agent after handoff.
• Remain the architectural owner of the agents you build, advising on changes that affect their structure and behavior.
• Review Agent Technician pull requests and lead refactors that require deeper architectural work.
• Assess whether expansion requests can reuse existing features or require new capabilities.
• Turn patterns across deployments into reusable components and patterns so that future agents are built faster

Skills

Conversational design

Voice user experience

Prompt engineering

Agent architecture

Cross-functional collaboration

Tools

APIs

Agent-building platforms

Job description

Hello Patient is hiring an AI Agent Product Manager with a strong background in building AI voice agents from scratch. You’ll design how our agents listen, interpret, respond, and guide patients through real healthcare workflows, turning complex requirements into conversations that feel natural and accomplish the task.

This hands-on role requires creating prompts, conversation flows, and agent configurations yourself, then using simulations and real chats to identify and improve issues.