Open-source structural solver and browser workspace.
Model, solve, inspect, and share 2D and 3D structures in the browser. The same structured model and solver can be used directly by engineers or driven through AI build/review workflows. Rust solver compiled to WASM. No installation required.
Try it now · Why it exists · Features · Getting started · Docs
3D industrial warehouse with Pratt roof trusses and crane bridge. Orange overlay shows the deformed shape under load. 216 nodes, 538 elements, 30 supports.
Same structure with stress utilization color map (σ/fy). Blue = lightly loaded, yellow = moderate, red = approaching yield.
The dominant structural analysis tools — SAP2000, ETABS, Robot, RFEM — cost thousands of dollars per year, run on Windows, require installation and license servers, and are closed source. Open-source solvers like OpenSees are powerful but require scripting and have no visual interface.
Stabileo is different:
- Browser-native. Open stabileo.com and start. No download, no license key, no account.
- Real solver. Rust engine compiled to WebAssembly — linear, nonlinear, dynamic, shells, staged construction, contact, fiber beams, and more.
- Real-time. The solver runs on every edit. Move a node, change a load, resize a section — results update instantly.
- Structured model surface. The browser UI, backend APIs, and AI workflows all target the same model/snapshot contract instead of hidden prompt magic.
- AI-ready, but deterministic. AI can help generate, edit, review, and explain models; the solver remains the source of truth for mechanics.
- Open source. Read the solver, trace the math, submit improvements.
- Transparent. Interactive step-by-step wizard shows every stage of the Direct Stiffness Method with KaTeX-rendered matrices.
Tech stack: Svelte 5 frontend, Rust solver engine via WASM, Three.js 3D visualization.
Originally built for structural engineering courses at FIUBA (University of Buenos Aires). Named after Daedalus, the architect who built wings to escape the labyrinth.
Stabileo's strongest technical wedge is not "AI chat" by itself. It is a structured structural model and a deterministic solver that humans and AI can both operate.
- Engineers can model directly in the browser and inspect diagrams, stresses, reactions, and diagnostics.
- AI workflows can build or edit the same structured model snapshot, then hand it to the same solver for real analysis.
- Review and explanation tools sit on top of solver artifacts and diagnostics instead of inventing mechanics.
Start here:
- 2D and 3D linear static, second-order, buckling, modal, response spectrum, time history, harmonic response, and moving loads
- Corotational and material nonlinear analysis, plastic analysis, fiber beam-column elements
- Staged construction, prestress/post-tension, cable analysis, contact/gap behavior, nonlinear SSI
- Initial imperfections, residual stress, creep/shrinkage
- Multi-family shell stack: MITC4 (ANS + EAS-7), MITC9, SHB8-ANS solid-shell, curved shells
- Guyan and Craig-Bampton model reduction
- Sparse-first assembly and solve with AMD ordering, 22-234× speedups on shell models
- Load combinations, envelopes, section analysis, stress recovery, kinematic diagnostics
| Code | Scope |
|---|---|
| AISC 360 | Steel |
| ACI 318 | Concrete |
| EN 1993-1-1 (EC3) | Steel |
| EN 1992-1-1 (EC2) | Concrete |
| CIRSOC 201 | Concrete |
| AISI S100 | Cold-formed steel |
| NDS | Timber |
| TMS 402 | Masonry |
| ASCE 7 / EN 1990 | Loads and combinations |
Benchmarked against NAFEMS, ANSYS Verification Manual, Code_Aster, SAP2000, OpenSees, Robot, STAAD.Pro, and textbook solutions. See BENCHMARKS.md for full coverage.
Use it now. Open stabileo.com. Works on any modern browser.
Run locally:
git clone https://github.com/lambdaclass/stabileo.git
cd stabileo/web
npm install
npm run dev # http://localhost:4000npm test # run the web test suite
npm run build # production build -> web/dist/Requires Node.js >= 18.
| Document | Contents |
|---|---|
| docs/README.md | Docs hub: quick start, AI workflow, solver reference, and roadmap entry points |
| QUICKSTART.md | First model tutorial: build, solve, inspect, and share a 2D beam |
| AI_MODELING_WORKFLOW.md | How AI build/review flows use the structured model + solver loop |
| SOLVER_REFERENCE.md | Coordinate conventions, model objects, outputs, and execution surfaces |
| SOLVER_ROADMAP.md | Solver status, sequencing, performance, and validation |
| PRODUCT_ROADMAP.md | App, workflow, and market sequencing |
| INFRASTRUCTURE_ROADMAP.md | Backend, deployment, auth, persistence, and operational sequencing |
| AI_ROADMAP.md | AI capability sequencing, safety rules, and prerequisites |
| BENCHMARKS.md | Validation coverage and benchmark status |
| VERIFICATION.md | Testing philosophy, fuzzing, invariants |
| POSITIONING.md | Market framing and competitive strategy |
| engine/README.md | Rust solver engine API and analysis types |
| CHANGELOG.md | Milestone updates |
| docs/research/ | Shell-family research, competitor comparisons, numerical methods |
Pull requests are welcome. For major changes, open an issue first to discuss the approach.
To report a vulnerability, email security@lambdaclass.com.
- Bautista Chesta — Civil Engineer (FIUBA), UX/UI and project management
- Diego Kingston — Ph.D. in Engineering (UBA), product–solver integration
- Federico Carrone — Founder of Lambda Class, solver lead
With contributions from mathematicians, physicists, computer engineers, and computer scientists at Lambda Class.
In honor of Daedalus, who built the labyrinth and dared to fly.