Best-practice workflows

Recipes that combine features into real work. Each is a pattern you can adopt wholesale or adapt.

The daily capture loop

Goal: never lose a thought; never spend time filing.

1. Capture freely all day — type quick notes, forward emails to your Inbox address, drop in meeting notes.
2. Let the autorouter file them; let extraction tag and link them.
3. Once a day, open Chat and ask "What did I capture today, and what needs a decision?" — triage from the answer, not from a pile of unread notes.

Not sure where to start? The TRY ASKING suggestions seed the daily triage question for you.

The meeting → decision → action pipeline

Goal: meetings that produce durable, findable outcomes.

1. During the meeting, write a note naming the people and the project in prose (that's what creates the entities and links).
2. Record decisions explicitly ("Decision: …") — the pipeline recognizes them, and clear phrasing helps later retrieval.
3. Write action items as natural sentences ("Priya to send the BOM by Friday"); confirm the important ones landed in the sidecar's Action Items, adding dates.
4. Later, ask Chat "What did we decide about \<topic>, and why?" — you'll get the decision plus a citation to the meeting it came from.

Asking Chat about a past decision returns the answer plus citation pills back to the meeting it came from.

The weekly review

Goal: a status you can trust, in five minutes.

1. In Chat: "Summarize this week's decisions and open risks for \<project>."
2. "What's on my plate?" — your owner identity makes this your items.
3. Save the good synthesis as a note (Chat → Save as note) so the review itself becomes a durable, searchable artifact.

Research synthesis

Goal: turn scattered reading into one coherent view.

1. Capture research notes as you go, tagging them (#research/\<topic\>).
2. Ask Chat to synthesize_topic — "Synthesize what I know about \<topic>." It reads across every relevant note and writes a cited summary.
3. Follow the citation pills to the sources to verify and go deeper.

Search complements Chat for research — sweep every relevant note directly, then follow results into the editor.

Working through your assistant (MCP)

Goal: your AI has your context, always.

1. Connect Claude Desktop / Code to your corpus.
2. Start a work session with "Using my Aletheia notes, catch me up on \<project> and list open items."
3. Have the assistant write back — draft a decision record or a summary and file it into your corpus with a note-creation tool. It's now part of your searchable knowledge base.

Everything these workflows produce lands back in the workspace — one filterable table over your whole corpus.

---

The meta-pattern

Capture in plain English → let extraction and the autorouter do the structuring → get it back by asking, not digging. Every workflow above is a variation on that loop. The discipline that pays off most is simply naming people and projects explicitly when you write — that single habit is what makes everything downstream connect.

Next: Troubleshooting & gotchas →