Handeia x interface forge - Generative UI Global Hackathon: Agentic Interfaces
AI Tinkerers - Durango, MX
Hackathon Showcase

Handeia x interface forge

Team consisting of a Bluver5 developer and UTD student skilled in Flutter/Dart, React, and integrating TensorFlow Lite and OpenAI APIs for hybrid AI.

1 member

Handeia — Adaptive Generative Workspace

Handeia is an AI-native adaptive workspace where interfaces are generated and transformed dynamically at runtime based on user intent.

Unlike traditional AI applications that return only text or static dashboards, Handeia treats the interface itself as a living system. The workspace continuously reorganizes, mutates, and adapts its visual structure depending on the context, task, and information requested by the user.

The project explores a core idea:

“Interfaces should not be static pages — they should evolve around human intent.”

Instead of navigating predefined screens, users interact with an intelligent runtime capable of transforming the workspace into:

analytics dashboards,

dependency graphs,

workflow systems,

approval flows,

executive summaries,

operational war rooms,

contextual visualizations, all generated dynamically from AI interpretation and structured context.

To demonstrate this concept within the hackathon timeframe, Handeia uses a simulated enterprise MCP (Model Context Protocol) layer that provides structured business context such as:

sales metrics,

events,

alerts,

predictions,

dependencies,

workflows,

approvals,

connected entities.

The AI interprets user intent and orchestrates runtime UI transformations using reusable adaptive interface primitives such as:

cards,

graphs,

flows,

nodes,

lists,

panels.

A key design principle of Handeia is:

“The UI is never created or destroyed — it transforms.”

This allows the workspace to maintain visual continuity while evolving into different operational modes in real time.

Examples of transformations include:

converting analytics into workflow views,

turning alerts into dependency graphs,

reorganizing layouts into executive summaries,

generating operational “war room” interfaces from business anomalies.

The system also includes an intelligent “seed state,” meaning the workspace is already active before the user submits a prompt. Rather than showing an empty screen, Handeia initializes a living contextual workspace that continuously adapts as interaction evolves.


Technologies Used

Core Frameworks

Next.js

React

TypeScript

Tailwind CSS

Generative UI Stack

CopilotKit

AG-UI protocol

MCP architecture concepts

A2UI-inspired adaptive rendering concepts

AI Models

Claude

Gemini

Runtime Architecture

Adaptive UI transformation engine

Runtime component mutation system

Intent-driven workspace orchestration

Structured MCP simulation layer

Interface System

Reusable adaptive primitives:

Cards

Graphs

Flows

Panels

Nodes

Lists

Data Layer

Mock MCP business context system

Structured JSON enterprise simulation

Event and relationship graph simulation

Additional Concepts Explored

Real-time adaptive rendering

Dynamic interface mutation

Spatial UI orchestration

Intent-driven visualization

Runtime-generated workflows

Handeia is designed as a first step toward a future where software behaves less like static applications and more like intelligent adaptive systems capable of reorganizing themselves around human goals and context.

AI Tinkerers Anthropic CopilotKit Google DeepMind Universidad Tecmilenio

Github

Summarizing URL...