Agent Product Manager (Remote - US East Coast)

Quadrivia is a health technology company building Q, a customizable AI assistant designed by clinicians to support healthcare professionals across the care continuum. The Agent Product Manager will own healthcare customer agents end to end, including on-site deployment, configuration, testing, production performance, clinical quality, and continuous improvement. The role combines forward-deployed customer work with product ownership and requires regular travel to US customer sites.

Responsibilities:

• Go on site. Sit with schedulers, care coordinators, practice managers and clinical leads, watch how the work actually happens, and translate that into agent behaviour. Expect regular US travel
• Wire agents into the customer's real environment: EHR data and workflows, phone systems and call routing, scheduling platforms, insurance and eligibility steps, clinical protocols. You will work alongside our integration engineers; you are the person who knows what the agent needs from that stack and why
• Be the AI and technical voice in the room — for the customer during deployment, and for Sales in demos, proofs of concept and scoping conversations
• Configure agents in Cortex (our agent platform) end to end: conversation graph (nodes, edges, transition conditions), prompts, variable extraction, escalation and confirm-then-act logic
• Build and maintain journeys — the orchestration around the agent: scheduling, retries, channel selection, call windows, escalation paths
• Define core KPIs and metrics that map to improving our customers' outcomes and goals
• Work with the broader product team improving the platform and localizing our tools to new markets and segments to drive automation
• Author durable regression suites in Starfish (our testing toolkit): scenarios, personas, judges and campaigns that cover an agent's real behaviour, not just the happy path
• Reproduce any production failure as an exact scenario, A/B the fix, and regression-gate the whole suite before anything ships
• Work with Clinical and Quality so that "passing" maps to clinical safety, not just to a green test. Clinical review is a gate on every behaviour change: nobody approves their own prompt change, including you
• Own the agents on your accounts for as long as they are live. When behaviour is wrong you find out why, fix it, and prove the fix. You close issues when the behaviour is right, not when it is explained
• Own the numbers that say whether the automation is working — containment, resolution, escalation rate, clinical quality flags — and be the person who reports them to the customer
• Separate an agent misconfiguration from a genuine platform bug, and hand the second one to Engineering with the evidence
• Carry roughly twenty agents across your accounts, not one. This role only works if you are building patterns and instantiating them, not hand-crafting every time
• Turn repeated manual work into a tooling requirement and push it to Product and Engineering — you are their most important internal customer
• Write the templates, standards and definition of done for agent quality that the rest of the company builds against

Requirements:

• • US healthcare fluency, from the inside. You have worked with or inside US provider organisations and you understand how a health system actually runs — scheduling and referrals, the front desk and the call centre, EHR workflows, eligibility and prior authorisation, and why clinical staff trust or distrust a new system. You know the difference between what a health system says it wants and how the work gets done on a Tuesday afternoon. This is not a preference. You will be in the room with practice managers, clinical leads and health system executives, and they decide in the first ten minutes whether you understand their world
• • US-based on the East Coast (Eastern timezone), and able to travel to customer sites in the US regularly. We need the overlap with both our customers and our European team, so Pacific-timezone candidates are not a fit for this seat
• • 3–6 years in a role that demanded fast mastery of an unfamiliar technical domain in front of a customer: forward deployed or solutions engineering, technical product, top-tier consulting, or an early-stage company where you were the person who figured things out and then shipped them
• • Demonstrated fluency with LLM systems in practice — prompt design, structured extraction, evaluation, and why agents fail. Opinions formed by having shipped something
• • Comfort reading logs, dashboards, traces and configuration files, and reasoning about a system from its telemetry. SQL and scripting are expected; writing production code is not required, though you are welcome to
• • Exceptional written communication. Most of your output is a specification, a test case, a root-cause explanation or a customer-facing report that someone else must act on
• • Rigour about correctness in a clinical context. You will change systems that talk to patients, and you are one of the last checks before that happens. You do not need a clinical background — we have clinicians and you will work with them — but you do need to take that responsibility seriously
• • Hands-on experience with Epic, Cerner/Oracle Health, athenahealth or similar, and with contact centre telephony
• • Experience with enterprise customers and stakeholder management
• • Conversational AI or voice-stack experience (ASR/TTS, LiveKit, Twilio or equivalent)
• • Experience building evaluation or regression-testing systems for non-deterministic software
• • A second working language, Spanish above all. Mandarin is also valuable