Fixed Screens vs Flexible
Two ways BranderUX turns an AI response into UI — choose by how much control you want.
Both modes are generally available and production-ready. You set the mode per project in the dashboard, and you can change it anytime without touching code. The choice affects who decides the layout, how much you set up in advance, and whether BranderUX's screen-selection step runs — but not how it streams: both modes paint a skeleton first, then fill in element-by-element as the data arrives. Either way, the end result is branded, interactive elements wired into the click-to-query loop.
Fixed Screens (Deterministic)
Your PM or designer builds screen layouts up front in the visual Screen Builder. On each query, BranderUX's AI scores all your screens and hands your AI the top few as tools — each with usage instructions and a fit hint — plus a plain-text fallback. Your AI picks the best-fitting tool and calls it with the data, and that screen renders.
- Layout decided by: you — pixel-level and consistent every time.
- Setup: build the screens once; reuse them across queries.
- Good fit when: you want a designed, predictable experience and PM/designer ownership of the UX.
Flexible (A2UI)
Instead of binding the AI to one predefined screen, BranderUX gives your AI the full element set plus the screens your PM/designer built as example layouts, and the AI composes the UI itself per query — choosing elements and arranging them in rows and columns. Your designers still shape the result through those examples; the AI just isn't limited to a single screen. It emits declarative A2UI messages that BranderUX renders.
- Layout decided by: the AI — it composes each response from your elements, guided by your PM/designer's screens as examples.
- Setup: none extra — your existing screens are reused as the examples the AI learns from.
- Good fit when: your queries are open-ended and varied, or you'd rather let the AI design each response than commit to one predefined screen.
Side by side
| Fixed Screens | Flexible (A2UI) | |
|---|---|---|
| Who designs the layout | Your PM/designer, up front in the Screen Builder | The AI per query — guided by your PM/designer's screens as examples |
| Consistency | High — the same screen every time | Varies with each response |
| Setup | Build screens once in the dashboard | None extra — your existing screens are reused as examples |
| Best for | Predictable, designer-controlled experiences | Open-ended, varied queries |
Filling the UI with real data
In both modes, BranderUX injects the current date and a “use real values, never fabricate” instruction into the prompt server-side — so time-relative queries resolve to today and the UI is less likely to render empty or with placeholder values. If your agent has its own data tools, have it fetch the data before it renders.
Switching modes
You can switch a project between modes anytime from the dashboard — no code change, no redeploy. Pick whichever fits the experience you want; you are not locked in. New to integration? See Choosing Your Integration and the Quick Start.