Senior Manager, Product Management

Salesforce Salesforce · Enterprise · San Francisco, CA

Salesforce is seeking a Senior Product Manager to own a core area of Agentforce Authoring, a platform for building, evaluating, and managing AI-powered agents. The role requires a hands-on approach, fluency with coding agents, and a track record of shipping products for technical users. The PM will own the roadmap, define success metrics, prototype ideas, dogfood the product, and collaborate across engineering, design, and other PMs to ship and iterate on agent authoring experiences.

What you'd actually do

  1. Own the roadmap, outcomes, and direction for a specific authoring surface (declarative builders, developer tooling, evaluation, or lifecycle).
  2. Prototype with coding agents to pressure-test ideas before writing a single spec. A working demo beats a 10-page doc.
  3. Dogfood Agentforce constantly. Build agents. Break them. Eval them. Bring the receipts back to the team.
  4. Integrate customer research, telemetry, and your own usage into what ships next.
  5. Coordinate with peer PMs across platform, runtime, evaluation, and client surfaces so the pieces fit together for customers.

Skills

Required

  • 5+ years in Product Management
  • Track record shipping products for technical users (developers, admins, or technical SMEs)
  • Fluent with modern AI tooling
  • daily user of coding agents
  • built and shipped agents yourself
  • substantively about LLM orchestration, evals, and what makes an agent actually work in production
  • Strong written and verbal communication
  • Sharp product taste

Nice to have

  • Fluency across both low-code and developer-first tools, with a sense for how they should connect
  • Technical depth: comfortable reading code, reviewing APIs, prototyping in a notebook, and partnering with engineers on real trade-offs
  • Deep familiarity with the current agent-building landscape (Cursor, Claude Code, Replit, Vercel, agent frameworks, eval tools) and a strong opinion on what they get right and wrong
  • Design sense: can turn messy systems into clean primitives with opinionated defaults

What the JD emphasized

  • fluent with coding agents
  • use them daily
  • prototype your ideas instead of describing them
  • eval your own agents
  • live inside the product you're shipping
  • technical users
  • daily user of coding agents
  • built and shipped agents yourself
  • LLM orchestration
  • evals
  • what makes an agent actually work in production
  • comfortable reading code
  • reviewing APIs
  • prototyping in a notebook
  • partnering with engineers on real trade-offs

Other signals

  • AI CRM
  • AI-powered agents
  • Agentforce platform
  • build, evaluate, and manage AI-powered agents
  • reason, act, and integrate deeply with CRM and external systems
  • author, evaluate, deploy, and refine AI agents end to end
  • technical users
  • daily user of coding agents
  • built and shipped agents yourself
  • LLM orchestration
  • evals
  • agent actually work in production