Principal Product Manager, Agents

LeanData•Santa Clara, CA

15h

About The Position

We are seeking a Principal Product Manager to own the product strategy and execution for our agentic products. This role involves taking ambiguous problems and resolving product, system behavior, and interaction models. You will work closely with RevOps and marketing ops leaders to understand their challenges and incorporate their feedback into the product. A key aspect of this role is defining the correct behavior for agents, building golden sets, and writing meaningful evaluations. You will also be responsible for designing trust mechanisms, including permissions, approvals, recovery paths, and audit trails, while navigating complex data systems like Salesforce.

Requirements

7-8+ years in product management, owning products end to end.
Experience taking at least one agentic product to launch in production.
Experience writing and debugging evaluations by reading traces.
Ability to reason about tool calls, context, retrieval, latency, and cost as product decisions.
Strong UX judgment with defensible opinions on interfaces.
Comfort working without extensive scaffolding, building necessary tools and instrumentation.
Genuine curiosity about B2B go-to-market operations and a drive to learn the domain deeply.

Nice To Haves

Experience taking a 0→1 product from concept to revenue in B2B SaaS.
Experience with the Salesforce ecosystem or other systems of record where governance and permissions are critical.

Responsibilities

Set direction and make decisions on product prioritization and scope.
Resolve ambiguous problems by defining product, system behavior, and interaction models.
Build prototypes and working examples to illustrate abstract ideas.
Collaborate directly with RevOps and marketing ops leaders to understand and address their challenges.
Drive design partner interactions to iterate on the product.
Define the correct behavior for agents on the product.
Build golden sets and write meaningful evaluations.
Analyze traces to understand agent behavior in live environments.
Characterize failures, identifying whether they stem from prompts, tools, context, or the product itself.
Design permissions and approvals to govern agent autonomy.
Design recovery paths for failed tool calls, incorrect data, or uncertain agent actions.
Design the audit trail for RevOps leaders to track agent actions and rationale.
Navigate integration with Salesforce and other data systems.