An open-source tool called VibeSolve takes plain-English descriptions of optimization problems and generates Timefold solver code — and its creators say it produces noticeably better results than prompting an LLM directly.
Mathematical optimization covers scheduling, logistics, transport routing, and similar constraint-heavy problems. Writing solver code for these by hand requires deep domain knowledge. VibeSolve sits between “describe what you want” and “here’s working Timefold code,” targeting the prototyping phase specifically.
What the project claims right now:
- Better output than raw LLM prompting for Timefold-specific code generation
- Rapid prototyping only — the team is explicit that it does not produce production-ready code
- Technical skills still required to use and evaluate the output
- Open source — PRs and feedback are actively solicited
The honest framing is worth noting. The team describes VibeSolve as an exploration of “where LLMs can add value in optimisation algorithm development, and where they get in the way.” That’s a more useful lens than the typical “AI writes all your code” pitch. Optimization solvers are a domain where constraint modeling, objective function design, and solver configuration interact in ways that trip up general-purpose LLMs. A tool fine-tuned to one solver framework (Timefold, in this case) has a real shot at producing coherent output where ChatGPT or Claude would hallucinate constraint annotations.
NOTE
Timefold is an open-source constraint solver (Java/Kotlin/Python) used for vehicle routing, employee scheduling, and similar NP-hard planning problems. If you’re not working in that space, VibeSolve won’t be relevant — but the approach of domain-scoped LLM code generation is worth watching.
The project launched on Hacker News with modest traction (7 points, no comments at time of writing). Early-stage, small community, no usage numbers disclosed.
VibeSolve is live at vibesolve.ai and open source. If you build scheduling or routing systems on Timefold, it’s a low-cost experiment — describe a problem in English, compare the output to what you’d write manually, and file issues where it breaks.