GenUI (Generative UI) is a paradigm shift in how a digital product's interface is built: instead of designing every possible screen up front, you define a system — rules, context and a component catalog — and an AI model composes the concrete interface in real time, adapted to each user's intent at each moment. The screen stops being a fixed artefact and becomes the outcome of a conversation between the user and the system.
The obvious objection is control: how do you stop a generative model from producing screens that are off-brand, inaccessible or plain broken? The answer is that in well-built GenUI the model never generates code or draws pixels. It acts as an orchestrator: it picks components from a pre-approved catalog — your design system turned into a vocabulary —, combines them according to explicit composition rules, and the result is validated against a schema before it ever reaches the screen. It's governed generation: the model's creativity operates inside a perimeter defined by your design team, not by the model.
The piece that has made this viable in production is Flutter's genui SDK — published by Flutter Labs, the Flutter team's labs group at Google — together with the open A2UI protocol. With genui, the generated interface travels from the agent to the app as streamed structured data: messages that create surfaces, update components or modify the data model, which Flutter renders natively. There are no webviews, no generated code executing on the device, and no performance penalty: they are your app's own widgets, composed dynamically. The SDK is model-agnostic — it works with Gemini, with Claude, or with any LLM you connect through your backend.
Why is Flutter the natural platform for GenUI? Because a single component catalog performs on iOS, Android, web and desktop from the same codebase, and because its composable widget system fits naturally with the idea of an interface described as data. For a company, that means the investment in the catalog — the expensive part of the project — pays off across every surface at once. And if your app is already built in Flutter, GenUI can be introduced surface by surface, with nothing rewritten: you start with one concrete flow (search, onboarding, support) and expand based on results.
At Dribba we've been shipping AI to production for years — autonomous agents, enterprise RAG, voice apps — and we are the only Spanish consultancy in Google's official Flutter Partners directory. That combination is exactly what a GenUI project demands: product judgement to decide which flows benefit from generation, senior Flutter engineering to build the catalog, and real experience operating LLMs in production with guardrails, evals and observability. If you want to explore what GenUI would do in your product, we'll show you on your case — not with slides.
