LeanData logo

Principal Product Manager, Agents

LeanData

Santa Clara, CA · HybridFull-timePosted 3 days agoStill listed 2 days ago

Most applications go out cold — see where you stand first. No sign-up to start.

Watch jobs like this.

At a glance

Compensation
No compensation found
Location
Santa Clara, CAHybrid
Schedule
Full-time
Work Authorization
Not specified

Olive lists jobs from US employers, including remote roles you can work from the United States.

Job overview

The Principal Product Manager, Agents will own product direction, resolve ambiguous problems, build prototypes, and collaborate closely with RevOps and marketing ops leaders. They will define agent behavior, design permissions, recovery paths, and audit trails while navigating Salesforce and other data systems to ensure trustworthy AI agent operations.

Skills & qualifications

RequiredNice to have

Skills

Product ManagementPrototype DevelopmentRevOps CollaborationDesign Partner InteractionAgent Behavior DefinitionTrace AnalysisFailure CharacterizationPermission DesignRecovery Path DesignAudit Trail DesignSalesforce Data Systems Navigation

Qualifications

7-8+ Years in Product ManagementAt Least One Agentic Product LaunchedWrote Evals YourselfDepth to Reason About Tool Calls, Context, Retrieval, Latency, and Cost as Product DecisionsReal UX JudgmentComfort Without ScaffoldingGenuine Curiosity About B2B Go-to-Market OperationsTaking a 0→1 Product From Concept to Revenue in B2B SaaSSalesforce Ecosystem or Any System of Record Where Governance and Permissions Are Non-Negotiable

Full job description

What You’ll Do

Own the Product

  • Set direction and make the calls on what ships, in what order, and what gets cut
  • Take ambiguous problems and resolve product, system behavior, and interaction model together, not sequentially
  • Build the prototypes and working examples that make an abstract idea arguable
  • Work directly and often with RevOps and marketing ops leaders, in their environments, so as to bring the reality of their challenges into solutions
  • Drive design partner interactions to iterate the product

Define What Good Looks Like

  • Define the correct behavior for the agents on this product
  • Build the golden sets and help write meaningful evals yourself
  • Read the traces. Build intuition for how the agent behaves on live work, not on your test cases
  • When it goes wrong, characterize the failure: prompt problem, tool problem, context problem, or product problem

Design for Trust

  • Design the permissions and approvals that define what the agent does on its own and what it must ask about
  • Design the recovery paths: a failed tool call, wrong data, an uncertain agent
  • Design the audit trail, so a RevOps leader can prove next quarter what the agent did and why
  • Navigate operating alongside Salesforce and other data systems, where the records live and the actions land

Qualifications Required Qualifications

  • 7-8+ years in product management, owning products end to end
  • At least one agentic product you took to launch. Not a prototype, not an internal pilot, not a feature with an LLM call in it. Something that ran in production and taught you things you’d do differently
  • You wrote evals yourself. You’ve built the suites, set the pass criteria, and debugged failures by reading traces. You know how far apart eval theory and eval practice are
  • Depth to reason about tool calls, context, retrieval, latency, and cost as product decisions, and to hold your own with senior engineers
  • Real UX judgment. Opinions about interfaces you can show the work behind and defend past the first objection
  • Comfort without scaffolding. There’s no eval harness, research cadence, or instrumentation waiting for you. You build what you need and move
  • Genuine curiosity about B2B go-to-market operations. Not a RevOps expert on day one, but energized to learn the domain deeply

Preferred Qualifications

  • Taking a 0→1 product from concept to revenue in B2B SaaS
  • The Salesforce ecosystem, or any system of record where governance and permissions are non-negotiable

What This Role Is Not

  • Not a platform role. You’re not building services or shared capabilities for other product teams
  • Not a people management role, but a senior independent contributor role.
  • Not forward-deployed, solutions, or program management for someone else’s roadmap
  • We’re not asking you to train models. You don’t need to be a data scientist or an ML engineer

Similar jobs, posted recently

Open roles like this one, listed in the last 30 days.

You've read the whole posting — now see how you match it.