Obsidian workflow.

Use Obsidian as your source of truth and Empirical as the layer that makes the durable parts reusable across AI tools. This is a local-first, deliberate workflow, not a blind vault sync.

Recommended model

Keep writing in Obsidian. Use Empirical for the subset of notes that should reliably follow you into ChatGPT, Claude, coding agents, and future workflows.

Fast setup path

Install Empirical, sign in, then start by recording a few high-value notes manually so the structure is right before you automate anything.

empirical install

# Then capture durable notes through your agent or API workflow:
record_graph_memory({
  category: "build",
  summary: "API auth uses OAuth bearer tokens with x-api-key fallback for V1 compatibility."
})

What to move over

Prioritize permanent notes, project decisions, architecture guidance, personal preferences, and reference material you repeatedly need in AI workflows.

Project constraints that would otherwise be re-explained every session
Durable preferences and conventions
Research notes that stay useful beyond one conversation

What to avoid

Do not dump the entire vault into Empirical. Daily scratch notes, transient journal entries, and private material that does not improve future AI work should stay in Obsidian only.

Current product status: There is no hosted one-click Obsidian connector in the current release. The useful path today is curation first, automation later.

Last updated: June 27, 2026