mylana tracks what you have, what you owe and what's coming, then gives you one short, number-grounded recommendation. Every figure is computed by tested code — the AI never touches the math. Nothing is uploaded, ever.
Pick your system, run it, and mylana opens in your browser. No account, no setup, no subscription.
Each card downloads the file directly. Or browse the full release on GitHub.
The macOS build is Apple Silicon (M1 and newer) only — on an Intel Mac it will not run. Check Apple menu → About This Mac: it must say Chip, not Processor: Intel. On an Intel Mac, build from source instead — it works fine, it just isn't prebuilt.
Install a newer version straight over the old one. Your data lives outside the app and the database upgrades itself on first run — you never lose a month, and you never have to export anything first.
Every number comes from pure, tested functions. The model only reads the finished figures and writes the advice — it cannot change them.
The LLM runs locally through Ollama. The only network call is an exchange-rate lookup, which sends currency codes — never your data.
Interest-bearing balance kept apart from interest-free months, resolved due dates, and a saved summary per month so you can see any past month at a glance.
A single self-contained binary — no Python, no Docker, no cloud account. It works offline.
mylana is free and open source. The code, the issues and every release live on GitHub — read exactly what it does, fork it, or open an issue. It's Go, so building it yourself is one command; the manual has the steps for every platform.