Decision Flow - Generative UI Global Hackathon: Agentic Interfaces
AI Tinkerers - Austin
Hackathon Showcase

Decision Flow

4 members Watch Demo

We built Decision Flow: an AI agent that takes one single user request (spoken via voice or typed, vague or highly technical) and, at runtime, dynamically generates the entire interactive decision UI — a clean, brief dynamic set of interactive decision components (including Yes/No buttons, 1–10 sliders, multiple-choice selectors, map inputs, and more). There is no chat window, no message bubbles, no ongoing conversation, and no pre-built dashboard or forms. After the user completes the micro-decisions in under 60 seconds, the agent immediately renders the final polished multimodal output with visuals, step-by-step guidance, and next actions.

- Gemini API — called directly via fetch - Google Fonts — Instrument Serif - Leaflet.js — for map widgets AI Tinkerers DM Mono Fraunces Full tech stack: Google DeepMind Inter Tight no SDK or abstraction layer

Codebase

Summarizing URL...