AI Product Manager

At Zopa, we're building AI Banking: an intelligent, conversational experience that lets customers do almost anything with their money just by asking. Move money, split a bill from a photo of a receipt, find a payment from last year, understand where their money went, freeze a card or set a savings goal. It is live to customers today and on its way to becoming the primary way people interact with Zopa.

Underneath it sits a platform we have built ourselves: an agentic loop, MCP-based tools, memory, a generic front end and an evals framework. Because that platform exists, the work has changed shape. A new customer experience is often a prompt change, a tool definition and a set of evals, not a quarter of engineering. One person with good judgement can now design, build and ship an entire experience.

That has created a role that does not really exist yet in most companies and sits between product and engineering. As an AI Product Manager, you will own a customer problem end to end and build the answer yourself on our platform, with AI doing most of the typing.We are a small team with a lot of autonomy, shipping fast in one of the most heavily regulated industries there is. If the idea of having your own agentic banking experience in front of hundreds of thousands of customers within weeks appeals to you, read on.

A day in the life:

• Own a customer problem end to end: identify the opportunity, decide what should exist, build it, ship it and learn from what happens next.
• Prototype your own ideas rather than writing a document for someone else to build. Here, we often build before we debate.
• Create complete customer experiences on our agentic platform, often through prompts, tools and configuration rather than new services.
• Design the evals that define what good looks like for your experience, using them to inform iteration and as a key part of the release decision.
• Work directly in the codebase, raise your own PRs for engineering review and partner with engineers when a problem needs deeper technical work.
• Shape how your agent performs in production, balancing context, cost per conversation, latency, model choice and routing.
• Review real customer conversations and turn what you learn into meaningful improvements, often within days rather than weeks.
• Lead your own analysis and research: interrogate the data, run the numbers, review transcripts and build the insight needed to make a decision.
• Make sound judgements on risk, compliance and potential customer harm as you build, working closely with Risk and Compliance partners.

About you:

You’ll likely bring many of the following:

• You are a builder. You can point to customer experiences you have personally built and shipped using AI, and clearly explain the decisions you made along the way.
• You have strong product taste. You know what good looks like, and just as importantly, what is not ready to ship.
• You can uncover meaningful customer insights and turn them into experiences that solve real problems in new and intuitive ways.
• You use coding agents regularly and have a thoughtful view on where they add value, where they fall short and how to use them well.
• You have designed evals and can explain how they helped you identify issues that traditional testing alone would have missed.
• You understand how LLM systems work in practice, including context management, tool design, prompting, non-determinism, and trade-offs around cost and latency.
• You think in systems. When you encounter a recurring problem, you look for a way to solve the underlying cause, not just the individual symptom.
• You are deeply curious, open-minded and comfortable changing your approach as the technology evolves.

Added bonus

• Banking, fintech, payments or another regulated environment.
• Building and operating agentic systems in production, with real users and meaningful consequences.
• Model routing, context engineering or cost optimisation at scale.
• Voice, image-based or other multimodal customer experiences.
• Side projects, open-source contributions or things you have built simply because you wanted to.

What this role is not:

• This is not a traditional product role where you write requirements and hand them to a delivery team. You will be expected to build, test and ship customer experiences yourself.
• This is not a people-management or large-scale programme-management role. The model is small teams with high autonomy and outsized impact.
• Research is an important part of the work, but it is always in service of getting something valuable into customers’ hands quickly.
• Your progression will be based on the quality and impact of what you create, rather than team size, reporting lines or the breadth of your remit.