AI Product Manager

Role & responsibilities

• Own the agent experience and roadmap define which customer problems are worth solving with agents, the target outcomes, and the path to get there. Translate brand and user needs into clear requirements and priorities. Lead across time zones and partner closely with engineering to ship.
• Build agents in Agentforce  configure agents in Agentforce Builder (the conversational build/test/refine workspace): define subagents (formerly “topics”), assemble actions, and connect them into reliable, goal-completing flows.
• Write and manage prompt templates — author, ground, version, and govern prompt templates as living assets.
• Define actions — specify the actions agents take to complete jobs, partnering with engineering on Apex/Flow logic where needed.
• Stand up the data foundation in Data Cloud (Data 360) — Partner with engineering to set up data spaces (“pods”), streams and Data Lake Objects (DLOs), harmonized Data Model Objects (DMOs), identity resolution, and retrievers so agents are grounded in accurate, governed data.
• Own trust, safety, and quality — set policies for accuracy, transparency, escalation, and fallback; build the evaluation harness and human-in-the-loop QA; measure deflection, containment, resolution, and CSAT.

Preferred candidate profile

• 8+ years in product management at a Lead level. Not all years need be as an AI PM, but you must bring mature PM craft: discovery, prioritization, stakeholder leadership, and shipping.
• Hands-on building customer-facing agents on Agentforce — personally built and shipped agents, not just specced them.
• Prompt template authoring and management — writing, grounding, versioning, and maintaining templates in Agentforce / Prompt Builder.
• Defining actions — defining and wiring the actions an agent uses to complete tasks.
• Data Cloud (Data 360) setup — data spaces/pods, ingestion (streams DLOs), harmonization into DMOs, identity resolution, and retrievers for grounding.
• Shipped and operated a production conversational or agentic product— evaluation, guardrails, hallucination handling, and human-in-the-loop quality.

Nice-to-have qualifications

• E-commerce / agentic shopping experience (discovery, search, recommendations, conversational shopping; path to purchase).
• Salesforce Agentforce Specialist certification (AI-201); AI Associate or Admin a plus.
• Agent Script (deterministic agent control) and Agentforce Voice experience.
• Apex / Flow literacy; Service Cloud / Commerce Cloud context; retrieval grounding (keyword, vector, hybrid).