Inspiration
BUILD is an AI-powered architectural workflow platform for turning a rough design idea into something an engineer can actually use.
It helps architects, designers, and real estate teams move through the full design pipeline in one place:
choose a project preset like a house, duplex, or hotel lay out a 2D blueprint on a grid let AI agents measure rooms and track materials or cost morph the design into a 3D preview export a clean handover sheet for engineering review The key idea is that BUILD is not just a drawing tool. It is an agent-driven system that understands the design process. The Layout Agent handles measurements and spatial organization. The Estimate Agent tracks rough cost and material impact. The handover step packages the result into a format that is easier to share with engineers and collaborators.
For the Microsoft Build 2026 / AgentCamp event, the project is positioned as a cloud-native demo of how AI agents can improve design communication. It maps naturally to:
Azure OpenAI for the agent intelligence Azure Cosmos DB for collaboration and live state sync Azure Web Apps for fast deployment What it does BUILD is an AI-powered architectural workflow demo that turns a rough building idea into a measurable 2D blueprint, a simple 3D preview, and an engineer-ready handover sheet. It helps users move through the design process step by step with AI agents supporting layout, estimation, and export.
How we built it We built it with React and Vite, using a step-by-step frontend flow instead of one crowded dashboard. The app is split into multiple files for clarity, with shared state handled through a React context provider. The UI uses Tailwind CSS and Lucide icons, and the blueprint area is rendered with SVG so the grid, rooms, and placed elements feel interactive.
Challenges we ran into The biggest challenge was making the demo feel clear instead of overloaded. We also had to fix a few UI issues, including a corrupted Step 2 screen and element deletion behavior. Another challenge was balancing realism and simplicity so the project still felt impressive but easy to explain in a hackathon setting.
Accomplishments that we're proud of We are proud that the app now has a clean guided workflow, visible AI-agent behavior, interactive blueprint placement, 2D-to-3D preview, and export flow. We also got the app to build successfully after the refactor, and the UX is now much easier for judges to follow.
What we learned We learned that hackathon demos need a strong story, not just features. Breaking the experience into simple screens made the product much easier to understand. We also learned how important it is to keep the frontend architecture modular so the app can grow without becoming messy.
What's next for BUILD Next, we would connect the mock agents to real AI services, add live collaboration, and support real project saving through Azure Cosmos DB. We would also improve the 3D rendering, make the export truly generate PDFs, and add more professional engineering handoff features.
Log in or sign up for Devpost to join the conversation.