# UNcontrolled

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Juan Quintero](https://cerebralvalley.ai/u/juanqu), [Yulian Stiven Bedoya Buritica](https://cerebralvalley.ai/u/ybedoyab), [Alejandro Feria Gonzalez](https://cerebralvalley.ai/u/AlexFG)
- **GitHub:** https://github.com/ybedoyab/drafty-meta
- **Demo video:** https://youtu.be/UynJO88R2LY
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4548

Drafty — From Natural Language to Verifiable Engineering Simulations

Engineering simulation is powerful, but traditionally requires multiple specialized tools and expertise in CAD modeling, meshing, physical assumptions, and finite element analysis. This makes simulation time-consuming and difficult to access during early-stage product design.

Drafty bridges that gap by combining Meta AI with a real engineering simulation workflow. Users describe an engineering problem in natural language, and an AI assistant helps identify the appropriate model, request missing dimensions, establish materials and boundary conditions, and prepare a structured engineering scenario.

After reviewing and approving the proposed design, users enter an interactive 3D Studio where they can edit geometry, assign materials, apply forces and supports, and execute real finite element simulations.

Behind the scenes, Drafty uses CadQuery for parametric CAD generation, Gmsh for quadratic tetrahedral meshing, and CalculiX for finite element analysis, with cloud execution through Modal. Results include displacement fields, von Mises stresses, preliminary safety factors, force-equilibrium checks, mesh-quality diagnostics, and downloadable engineering reports.

Supported examples include L-brackets, cantilever beams, rectangular plates, perforated plates, and preliminary beam-element analysis of simply supported structures. Drafty also supports iterative thickness optimization to explore lighter designs under engineering constraints.

Our production demonstration validates a complete workflow: Meta Muse Spark interprets an aluminum L-bracket problem, the user confirms a 200 N load and support conditions, and Drafty generates the CAD model and solves a real C3D10 finite element mesh containing 18,980 elements and 31,795 nodes.

Drafty combines conversational AI with deterministic physics. The AI helps formulate and orchestrate the problem, while established numerical tools calculate and validate the engineering results. Human approval, traceability, and explicit limitations remain central to the workflow.

Live application: https://drafty-web-umber.vercel.app

User: danielossa556@gmail.com
Pass: test_meta

## More from Meta Global AI Developer Hackathon

- [Cumple&Cobra](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4545)
- [Lumen AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4546)
- [synapse](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4547)
- [R Majestic Studio](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4549)
- [Rahul Sah](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4550)
- [Fuad Hasan](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4551)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4548. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt