Obsidian
Local workflowBest fit today. Keep your vault where it is, decide what is durable, and move the right notes into Empirical on purpose.
Bring your notes
If you already keep a second brain in Obsidian, Logseq, Markdown, or Notion, the right move is not starting over. The right move is turning the durable parts of that system into reusable AI context.
Setup is gated behind signup. You can read the plan here, but actual workflow links stay locked until you have an Empirical account.
What this page does
Watch the flow
This demo is intentionally controlled. It shows the exact story we need: note capture, memory extraction, chat payoff, and coding payoff, without pretending the note sources are already wired into a live connector.
Video spec
`9:16`, `1080x1920`, `30s`
Beat structure
Obsidian -> Empirical -> ChatGPT -> CLI -> end card
Notes can come from anywhere.
selected from your notes
Deadline moved to Thursday, need to cut two slides from the deck.
Staged walkthrough
Short vertical concept demo for notes -> Empirical -> ChatGPT -> CLI
Best pathway
This is the right scope. A clean setup hub plus source-specific guides gives users a usable path immediately. Hosted connectors can come after we prove which workflows people actually use and which parts should be automated.
Current rule
Only Obsidian, Logseq, plain Markdown, and a manual Notion path should be live here today. Apple Notes and Google Docs should stay marked as coming soon until there is a workflow worth shipping.
Available now
Best fit today. Keep your vault where it is, decide what is durable, and move the right notes into Empirical on purpose.
Useful if your graph already contains project decisions, research, and reusable context you want available in AI tools.
The cleanest lowest-friction path. Works well if your notes already live in files and you want a simple repeatable routine.
Supported as a deliberate export-and-curate path for now. No one-click hosted sync yet.
Coming soon
Worth supporting, but not practical to promise until we have a clean extraction and refresh model.
Coming later. We need a workflow that is actually useful, not just an API connection that creates more cleanup work.
Read first
The blog post frames the problem correctly: people who already keep notes are the exact users who should not have to re-explain themselves to every AI tool.
Open the blogSetup docs
Source-specific docs let us explain the workflow clearly without bloating the landing page or faking product depth.
Browse docsProduct stance
The first useful version is a guided workflow hub. Once those paths are proven, we can automate the parts that are repetitive instead of automating the wrong thing first.