Cara Baestlein

At Dojo, a major UK payment provider, AI has to be right. As teams began building agents on top of their data, one thing became clear: an agent is only ever as good as the context behind it. In payments, "roughly right" isn't a category that exists. Consistency and accuracy are the name of the game. Like most companies of its scale, Dojo's data estate grew fast and grew across several systems, each excellent at its own job, but with no shared view or control across them. That puts a ceiling on what agents can safely do. Without a common, governed view of the estate, an agent can't reliably know what it's reasoning over, whether the data is fresh, who owns it, or whether it should have access at all. Solving that became a prerequisite for taking AI beyond experiments. In this session, Dojo shares how they built that context layer: clear ownership, proactive governance, and discovery across systems, so both engineers and agents can find data and trust what they find. They'll cover why open-source tooling and open governance mattered for avoiding lock-in, what worked, and the lessons worth passing on. We'll also show how the flexibility of the Aiven Data Cloud supports new use cases across clouds, from discovery to production AI running next to the data.