Keep the useful parts close

The decision you made last month. The constraint buried in a README. The explanation in an email. Serenity brings these sources into a personal brain so you can find them again without remembering where they came from.

Start with local files, add repository documentation, or connect a Gmail inbox. You choose the source scope and when to sync.

Ask, then check the source

With models configured, Serenity extracts claims and answers questions with citations. Follow the evidence back to the source instead of treating a generated answer as a new authority. Full-text search also works on its own, with no model.

Memory you can keep

Your brain lives in a Git repository you own. Its records are inspectable files. You choose the backup location and model providers. Model-assisted features send relevant source or query context to the provider you configure.

Bring your constraints into the conversation

Serenity also checks structured actions against recorded constraints, called precepts. A result can explain a violation and the recorded reason behind it. Free-text checking is still unverified, so use the documented structured-action contract.

Built for a deliberate first step

Serenity is for people comfortable with a terminal who want a personal, inspectable memory across their work. It is under active development. Start with local search, verify the results, then add model-assisted answers as useful.

The person behind the project

Good AI needs
thoughtful engineering.

I’m David Ndungu. I build software that makes AI useful in real work. Serenity is part of that work: keeping context available, decisions inspectable, and people in control.

Through my consultancy, I build custom AI automations for business workflows, connected to real tools with human approval gates.

Work with David ↗
Ask about Serenity