SLSepiaLog
Understand · SepiaLog feature

Local AI research assistant grounded in the project’s own history

Ask questions about research notes, files, metadata and history using a model that can run on the researcher’s own computer.

What is a local AI research assistant?

A local AI research assistant processes project context with a model running on the researcher’s computer rather than silently sending research files to a third-party service. SepiaLog grounds answers in local notes, metadata, files and history and shows the supporting records.

The research problem

Generic chat tools can produce fluent answers without knowing which project version, decision or note supports them. Sensitive research also raises legitimate questions about where prompts and source material are processed.

Research actionConnected documentationRecoverable output

How the workflow works

  1. Connect an optional local model through Ollama.
  2. Ask a project-specific question such as why a year was excluded.
  3. Review the answer together with the notes on which it is based.

What SepiaLog provides

  • Questions grounded in the local project
  • Source notes surfaced with answers
  • Draft README, methods and data-availability text
  • Supervisor updates based on recorded changes

Common questions

Is a local model required to search notes?

No. SepiaLog can still surface relevant notes without a model; generative drafting requires an optional configured model.

Does local automatically mean ethically approved?

No. Consent, lawful processing, access controls and institutional guidance still apply.

Can the answer be checked?

Yes. Supporting notes are shown so the researcher can verify the response.