Technical Product Manager, AI Agents

Hello!

Please give this Job Description a good application of your synapses. It's most certainly not for everyone. Note that 3+ years of technical (directly working with engineering) experience shipping products in messy environments where requirements weren't handed to you, and agentic builds are MUST HAVE experiences. Proximity to US healthcare workflows (prescribing, prior auth, coverage, specialty pharmacy, patient support) borders on MUST HAVE (if lacking, you'll really need to convince us you can dent an absurdly complex ecosystem). 100% remote though you must reside Stateside. Sadly, we can't facilitate VISAs. Cover letters are welcome, appreciated, and reviewed by a human.

About PrescriberPoint

PrescriberPoint, a Series A startup incubated by BCG Digital Ventures and funded by Lilly, Pfizer, MasterCard, and Adobe, is modernizing how U.S. healthcare professionals (HCPs) get patients on therapy. Our platform streamlines prescribing by bringing together trusted drug information, coverage and prior authorization requirements, patient affordability options, and practice tools—while enabling compliant collaboration with pharmaceutical brand teams.

Why this role exists

We’re building AI agents that do real work for healthcare teams—reducing friction in prior auth, prescribing support, and our marketplace workflows. We need a hands-on product leader who can turn messy workflows and ambiguous goals into shipped agent capabilities that users actually trust and use.

This is not an engineering role—but it is a deeply technical, highly accountable role. If you join, you’ll be measured by what ships, adoption, quality, and the business impact those agents deliver.

What You’ll Own (for Real)

• Clear user + business problem definitions
• Agent capability specs that engineers can build
• Quality + safety standards that prevent trust-breaking failures
• Release plans that drive adoption and measurable value
• Define, track, and communicate whether an agent is “successful” in production using success metrics (task success, accuracy, escalation rate, user trust signals).

If something isn’t working (accuracy, latency, UX confusion, stakeholder misalignment), you won’t “escalate” it—you’ll drive it to ground.

What You’ll Do Day To Day

Find the wedge + ship

• Live in the workflow: talk to users, watch sessions, read support tickets, and find the highest-leverage friction points.
• Turn those into a tight sequence of releases: MVP → v1 → v2 (not a PowerPoint roadmap).

Be the agent-product glue

• Translate between what models/agent SDKs can do (Claude Agent SDK, OpenAI Agent SDK, tool use, planning, memory patterns, evaluations) and what users need.
• Write specs that are unambiguous: inputs, tools, permissions, failure modes, confidence cues, human handoffs, and success criteria.

Own quality like it’s your name on it

• Define “good” in ways we can test: task success rate, groundedness, hallucination rate, safety triggers, escalation rate, and time-to-complete.
• Partner with engineering to build evals, test suites, and monitoring—then use them to decide what ships.

Drive adoption (not just delivery)

• Work with sales, marketing, and customer care to package each release into customer language: what changed, why it matters, when to use it, and how to trust it.
• Make sure outbound messaging reflects real capability and sets correct expectations (no vapor).

Operate cross-functionally without becoming a bottleneck

• Run lightweight rituals that accelerate shipping (fast decisions, tight feedback loops), not slow it down.
• Your job is to increase throughput and quality, not to be a “required approval step.”

What We’re Looking For

Builder DNA

• You’ve shipped products in messy environments where requirements weren’t handed to you.
• You can go from problem → spec → shipped feature → iteration without waiting on “perfect clarity.”

Technical fluency (without being an engineer)

• You can hold your own with engineers on agentic patterns and tradeoffs: tool calling, retrieval, orchestration, evaluation design, latency/cost, guardrails.
• You can explain those tradeoffs to non-technical audiences in plain language.

Bias to measurable outcomes

• You’re comfortable being measured by adoption, task success, quality, and business impact—not just “delivery.”

Healthcare Exposure (preferred)

• You’ve been close to US healthcare workflows (prescribing, prior auth, coverage, specialty pharmacy, patient support) or can learn fast and ask sharp questions.

What Success Looks Like (first 90 Days)

• You’ve mapped the most valuable workflows and identified 2–3 wedge agent opportunities.
• You’ve shipped at least one meaningful agent improvement to production (or a tightly scoped beta) with measurable success criteria.
• You’ve established the basics of a quality loop: evaluation plan, monitoring, and a release checklist that improves speed and trust.

What Success Looks Like (6–12 Months)

• Agents are delivering measurable reductions in time-to-complete and support burden, with increasing adoption.
• We can reliably ship improvements because evals/monitoring guard quality.
• Customers can clearly articulate the value—and product claims match reality.

This role is NOT for you if…

• You prefer coordinating over building. (This job is measured by shipped outcomes, not meeting output.)
• You need a fully baked roadmap and requirements handed to you. You’re expected to find the wedge, define the path, and drive it.
• You’re uncomfortable being personally accountable for adoption, quality, and impact after launch.
• You default to “aligning stakeholders” instead of making calls, writing crisp specs, and unblocking delivery.
• You want to be the gatekeeper for decisions or approvals. Your job is to increase throughput, not rate-limit it.
• You aren’t excited to dig into messy details: edge cases, failure modes, eval results, support tickets, and real user workflows.
• You’re looking for an AI role that’s mostly strategy/vision. This is hands-on: discovery → ship → measure → iterate.
• You can’t (or don’t want to) get fluent enough in agentic tech (e.g., Claude/OpenAI agent SDKs, tool use, orchestration, evals) to partner tightly with engineers and explain tradeoffs to customers.

Qualifications

• 3+ years product management in a technical facing role (working with engineering directly)
• Past experience in software development, a CS or related degree, etc. is preferred
• Healthcare (per the intro paragraph) and start up experience are highly desirable
• Strong versions of the personal attributes listed above

Compensation

Compensation ranges from $120,000 - $140,000 depending on location and experience. There is a bonus.

Why join PrescriberPoint?

• We have a really good shot at improving the millions of lives and careers of HCPs, Patients, and their families (even pets!)
• We hire adults with a Trust-first/It's All Life philosophy
• We have some great benefits for a firm at our stage: 401(k) w/matching, all kinds of insurance (including matching HSA and pets!), commute from your kitchen, Open PTO (which leaders use!), remote stipend, yearly education budget, and working with some of the smartest yet humblest and respectful people in the business
• We’re (objectively) way better looking than our competitors :-)

Key skills/competency

• AI Agents
• Product Management
• Healthcare Workflows
• Prior Authorization
• Product Shipping
• Technical Fluency
• Stakeholder Management
• Evaluation Design
• User Adoption
• LLMs