Urbanis AI

The Global Housing Crisis:* Nearly 40% of the world's population faces unaffordable prices, chronic shortages, and informal settlements, with millions more residents expected in cities by 2050.

Bureaucratic & Material Bottlenecks: Traditional construction is bogged down by slow, contradictory municipal regulations and heavily relies on carbon-heavy default materials like concrete and steel.

## What it does

AI-Orchestrated Ecosystem: A vertical integration platform connecting architects, public entities, and private organizations to deliver modular, carbon-negative, and seismic-resistant housing.
Six Specialized Agents: Uses dedicated AI agents (THOT, KAEL, IRIS, CRONOS, TERRA, LYRA) to automate everything from context gathering and regulatory code checks to live budgeting, sustainability tracking, and final documentation.

## How we built it

Multi-Agent Architecture (Gemini-based): Built with specialized agents that query specific regulatory texts, cost indices, and material databases directly rather than relying solely on generic model knowledge.
Deterministic Validation Layer: Kept safety and legal guardrails (like geofencing flood zones and protected land) in separate, rule-based code that the AI cannot override.
Tech Stack: Developed using a FastAPI backend and a React frontend (iso.html for modular design studio and predio.html for spatial/geospatial validation).

## Challenges we ran into

Regulating LLM Outputs: Ensuring that legally binding regulatory references and building codes were accurate and auditable without allowing generative "hallucinations."
Data-Gap Handling: Managing missing local institutional or municipal data dynamically so the system flags gaps rather than guessing or stalling the project workflow.

## Accomplishments that we're proud of

Comprehensive Workflow Integration: Successfully turning fragmented municipal research, spatial validation, material sourcing, and carbon tracking into one continuous, verifiable pipeline.
Democratizing Enterprise Tools: Equipping independent architects and small design firms with compliance and costing tools usually restricted to large developers.

## What we learned

Domain-Grounded vs. General Purpose: The critical importance of separating creative design synthesis from strict, deterministic rule-based safety checks (like environmental and flood vetoes).
Local Circular Economy Impact: How routing construction spending directly to regional eco-suppliers and local builders can drastically lower carbon footprints while keeping economic value in the community.

## What's next for Urbanis AI

Phase 1 & 2: Deploying pilots in partner cities to validate agent accuracy against real municipal approvals, followed by expanding KAEL's regulatory database to new states and countries.
Phase 3 to 5: Integrating a live local sourcing marketplace, adding predictive climate-risk modeling, and scaling the platform globally for UN-Habitat-aligned housing initiatives.