AI Agent Product Manager

Wonderschool; San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Job summary

Wonderschool is seeking an AI Agent Product Manager to own the systems and operating performance of AI agents across provider, parent, government, and internal workflows. You will identify valuable work for agents to do, design how agents use tools and work with people, and improve the quality, reliability, speed, and cost of those systems in production.

This is a hands-on product role based in San Francisco.

Qualifications

• You have shipped and improved software products or operational systems used by real people.
• You can reason about agent behavior, APIs, data, evaluations, and system tradeoffs with engineers, and you are willing to prototype hands-on.
• You use evidence from users and production performance to make product decisions.
• You communicate clearly across engineering, customer-facing, and operational teams.
• You are comfortable working in ambiguity and taking ownership from problem discovery through results.
• You care about child care providers, families, and the people who serve them. A particular degree is not required.

Responsibilities

• Map customer and internal workflows to find where an agent can complete a useful task or help a person do it faster.
• Define each agent’s job, tool access, decision boundaries, escalation path, and success measures.
• Partner with engineering, design, and operations to prototype, launch, and improve agent systems.
• Build evaluations and review real failures, customer feedback, and operational data to understand where agents succeed or struggle.
• Track quality, reliability, latency, cost, adoption, and customer outcomes; use those signals to prioritize improvements.
• Design clear human review and handoff experiences for cases an agent cannot or should not complete alone.
• Stay close to customers and frontline teams, and own the feedback loop from discovery through production performance.

Job description

Wonderschool’s mission is to ensure every child has access to early education that helps them realize their full potential. We help child care providers start, operate, and grow their businesses with software, coaching, and support. We also partner with governments to modernize public child care systems. Our small, in-person team is motivated by the impact we have on families and communities.

About the Role

We are looking for an AI Agent Product Manager to own the systems and operating performance of AI agents at Wonderschool. You will identify valuable work for agents to do across provider, parent, government, and internal workflows; design how agents use tools and work with people; and improve the quality, reliability, speed, and cost of those systems in production. You will work directly with customers and operational teams to understand the work, then partner closely with engineering to ship, measure, and improve agent experiences.

This is a hands-on product role based in San Francisco. You will be accountable for real outcomes, including work completed accurately, timely human handoffs, less manual effort, and better customer experiences.

What You’ll Do

• Map customer and internal workflows to find where an agent can complete a useful task or help a person do it faster.
• Define each agent’s job, tool access, decision boundaries, escalation path, and success measures.
• Partner with engineering, design, and operations to prototype, launch, and improve agent systems.
• Build evaluations and review real failures, customer feedback, and operational data to understand where agents succeed or struggle.
• Track quality, reliability, latency, cost, adoption, and customer outcomes; use those signals to prioritize improvements.
• Design clear human review and handoff experiences for cases an agent cannot or should not complete alone.
• Stay close to customers and frontline teams, and own the feedback loop from discovery through production performance.

What Success Looks Like

• Agents complete well-defined work accurately and consistently, with failures visible and addressed quickly.
• Customers and teams receive timely help and know when a person has taken over.
• Manual work and time to resolution decline while customer experience improves.
• The team can see how each agent performs, what it costs to operate, and where to improve it next.

Who You Are

• You have shipped and improved software products or operational systems used by real people.
• You can reason about agent behavior, APIs, data, evaluations, and system tradeoffs with engineers, and you are willing to prototype hands-on.
• You use evidence from users and production performance to make product decisions.
• You communicate clearly across engineering, customer-facing, and operational teams.
• You are comfortable working in ambiguity and taking ownership from problem discovery through results.
• You care about child care providers, families, and the people who serve them. A particular degree is not required.

Location

• San Francisco, California — in person at our Rincon Hill office, typically 5–6 days a week.

Why Join Us

• Build agent systems for consequential, real-world child care workflows.
• Work closely with engineering, operations, customers, and company leadership.
• Own a product area from first deployment through measurable operating performance.
• Join a small team where you can ship quickly and see the impact of your work.

How You’ll Work

• Use agentic tools to accelerate research, prototyping, testing, and iteration while checking their output against real customer needs.
• Spend time with customers and operational teams, including occasional evening or weekend events when the work calls for it.