We build custom software for small and medium businesses — internal tools, automations, and the occasional full MVP. Every project ships the same way: a plain-language description of what we built, plus the complete source code. Your software is yours. No lock-in.
Favur is our flagship project: an autonomous multi-agent software team. Hand it a Statement of Work and it plans, architects, codes, reviews, tests, and ships a complete, tested repository.
No favor too large.
Our favorite part isn't the finished repo — it's the replay. Repos in this org that Favur built autonomously say so at the top of their README, with a link to the full replay of the run that produced them: the whole journey from Statement of Work to shipped, tested code. Don't take our word for how it works. Watch it work.
Favur isn't publicly launched yet. Its home will be favur.dev.
We also run Favur Evals: an independent, self-funded public benchmark. The setup is simple — identical Statements of Work run through different LLMs inside Favur's own agentic pipeline, scored across the whole software development lifecycle. Eight axes: code quality, test quality, cost efficiency, tool discipline, workflow & reporting, effort efficiency, process discipline, and deliverables.
No model vendor sponsors it, and that's the point — every model on the board is doing something genuinely hard, and the scores say what they say.
- 🏢 awesoftsolutions.com — the company
- 🤖 favur.dev — Favur
- 📊 evals.favur.dev — Favur Evals