Tana

An AI-native note-taking workspace: an outliner over a graph data model where every bullet is a node, and supertags turn nodes into typed objects with fields. Introduced to this wiki by tana-supertags-guide.

What it is (from the source)

• Supertags = typed templates attached to a node (#Task, #Meeting, #Person, #Project, …), each with its own fields, supporting inheritance from base types — “a task can be a Supertag with fields for due date, priority, project, time estimate,” not just text + a checkbox.
• Everything is a node; nodes reference each other; Live Queries auto-update.
• AI-native: supports GPT-5.1, Gemini 3 Pro, Claude Sonnet 4.5; the AI uses your structured graph as context, transcribes meetings, and auto-tags/files voice memos.

The synthesis it represents

Tana ties together threads this wiki has been tracing separately:

1. Links + structure. It fuses the networked-nodes lineage of roam-research with notion‘s structured fields — the two poles of the earlier links-vs-databases contrast, combined.
2. Typed notes, again. Supertags-as-typed-objects are a third independent convergence on typed pages, alongside gbrain‘s schema packs and this wiki’s schema.org @type model — evidence that typing your notes is a recurring good idea.
3. AI over your structured graph. “The AI reads your graph, not a blank document” is precisely the llm-wiki / gbrain thesis — here arriving inside a mainstream PKM tool. Tana is where the tools-for-thought lineage and the LLM-wiki thesis meet.

Related

tools-for-thought · tana-supertags-guide · roam-research · notion · logseq · knowledge-graph · gbrain · llm-wiki

Re-verified 2026-08-04 — the product moved, and the wiki’s use of it did not

Homepage read directly at 67 days past window (this page had no url: until today, so nothing had ever been re-checked against a live source). Tana now describes itself as “an agentic meeting platform” — AI agents working during native video calls, creating documents, filing issues and updating context while people talk. Deployment is web plus Mac/Windows/Linux desktop, iOS and Android “coming soon”, on a Free/Pro/Team/Business freemium ladder with GitHub, Slack, Linear, Jira and HubSpot integrations.

Recorded, not overwritten. The description above — outliner over a graph, supertags as typed objects, live queries — came from tana-supertags-guide and was accurate as of that source. Whether the graph and supertag machinery still sits underneath the meeting product, or has been de-emphasised in favour of it, this page cannot say: a homepage tagline is positioning, not architecture, and the corpus holds no second source on the change.

Why it matters beyond this page. synthesis uses Tana as the point where “the lineage closes on itself” — the AI-native PKM tool that reads your typed graph instead of a blank document, and therefore the arrival of the llm-wiki thesis inside a mainstream product. If the company has pivoted from structured personal knowledge to meetings, that reading is at best dated and at worst was a snapshot of a waypoint rather than a destination. Flagged there as an open question rather than rewritten here, because one homepage read is not enough to retire a synthesis thread.