Tana builds a real typed graph — after you build the supertags. Get the same map with zero schema effort.
Tana’s context graph is genuinely good, but its magic is gated behind supertags and fields you define first — the famous learning curve, and the exact part people abandon. Naumu reaches the same kind of typed, connected map by reading your words and proposing the structure itself. And your own Claude and ChatGPT read and write that brain over MCP — something Tana’s in-app AI doesn’t do.
Why it holds up
A typed graph with no setup — that your own AI can write.
A typed graph, zero schema effort
Tana’s magic waits behind supertags and fields you define first. Naumu reaches the same typed, connected map by reading your words — it proposes the entity types and links itself. A first pass you correct in a glance, not a system you build before it pays off.
Your AI reads and writes it, both ways
Tana’s agentic AI lives inside Tana. Naumu exposes the graph over MCP: your ChatGPT writes a decision today, Claude knows it tomorrow — as you, with your exact permissions, and reversible. This is the seam I haven’t seen any tool in the category return.
Bring your Tana data — don’t port a structure
No connector, and you don’t need one. Export your nodes, drag the files in, and Naumu does a first pass you correct in a glance. You arrive full, not at a blank outliner — and you never re-create the supertags.
The honest comparison
Where each tool actually wins.
No strawmen. We both have a real graph — the rows where Tana wins are in here on purpose.
Schema effort to get there
Real — supertags and fields you define first; the learning curve everyone warns about.
Zero — it reads your words and proposes the typed structure; nothing to configure.
Capture
Pick a supertag, fill its fields, learn the outliner before the payoff lands.
Dump it raw — type, paste, drag, talk. It types and connects it after, no supertag chosen.
A real typed, connected graph
Yes — the context graph is genuinely strong, entities and relations you can traverse.
Yes — entities and relationships you can open and traverse, built from one ugly paste.
Two-way write-back from your own AI
Agentic AI lives inside Tana; Claude and ChatGPT can’t write back into your graph as you.
Your ChatGPT writes the same brain Claude reads — over MCP, as you, directly and reversibly.
Where a fact came from
Graph-grounded answers; the trace is there.
Every answer carries a receipt — the source and timestamp attached. A genuine tie here.
Staying organized over time
Supertag hygiene to keep up; the structure degrades the week you get busy.
No supertag debt to fall behind on — you never re-file it; it stays organized without upkeep.
Live in-meeting action (agents act during the call)
Best-in-class — the flagship ships outcomes live while you’re in the meeting.
Ask-it-later, not act-in-the-moment; Naumu structures after, it doesn’t run the call.
Hand-crafted structure for power users
Best-in-class — if you enjoy authoring rigorous structure, supertags go deeper with more control.
Programmable, but optimized for zero-effort first — control is an upside you reach for later.
Outliner & editing-surface maturity
Years of polish on the outliner, views, and automation.
A younger product; not built to be a deep editing surface.
Trying it
Sign up and climb the supertag curve before the graph is useful.
No signup to try; nothing saved until you keep, so the trial costs you nothing.
If your knowledge is born in meetings and you want agents shipping outcomes live in the call, or you genuinely enjoy authoring a rigorous structure and want maximum control — use Tana. It’s the better buy for that, and I’d rather tell you than lose your trust on a rigged row. Naumu is for the other reader: the one whose knowledge is born scattered across docs, chats, and tabs, who wants the typed map without the second job, and who wants Claude and ChatGPT to share one brain.
FAQ
Switching from Tana, honestly.
Compare Naumu with other tools
The map without the second job.
Export your Tana nodes and drag them in, paste a blob, or start with a sentence. Naumu types it, connects it, and from then on your Claude and ChatGPT share one brain.