What this is.
Tatara is a governed document store for AI agents and humans. A company writes down what it knows in one place (ideal customers, pricing, processes, playbooks), and every agent it runs reads from there instead of starting each conversation from zero.
Why Tatara exists
Because companies re-explain themselves to every AI they use, and the explanation dies when the window closes.
Anyone running agents on their own material knows the pattern. You paste the background into a chat, then paste it again tomorrow into a different tool. The day you forget, the agent answers confidently and wrongly, because it never had the context and has no way to know that it is missing. Nothing accumulates.
A company brain is where it does. The team writes down what the company knows in one place, and every agent reads the same current text instead of a paraphrase or a stale copy.
How agents connect
Over the Model Context Protocol (MCP). Claude, ChatGPT and Gemini already speak it, so you add a brain once in each tool and all of them read from it. There is no per-agent integration to build, and agents can connect on every plan, including the free one.
What “governed” means here
Every agent write into a brain must carry a type, a title, a description and tags, steered to that company’s own taxonomy. Every save is also stamped with who, or what, made it.
The stamp matters most. An agent cannot pass itself off as a person, so the record always says which writes came from your team and which an agent made. A write that does not fit is refused with an error naming the problem, and the agent corrects itself instead of filing something malformed.
That is the difference between a brain and a folder of notes. A document with no type, no description and no place in the structure is not a source of truth; it is something somebody left lying around.
Why the documents are plain Markdown
Because knowledge locked in someone else’s format is not really yours.
A brain is plain Markdown with YAML frontmatter: one file per concept, with the links between them forming a graph. There is no proprietary schema and nothing you could not open in a text editor. It is the same open shape Google’s Open Knowledge Format now formalises as a standard.
The reason is commercial rather than aesthetic. People should stay with Tatara because it is good, not because leaving is expensive.
Who builds Tatara.
I’m Angus. I built Tatara because I was pasting the same company context into every AI chat, and watching good agents produce confidently wrong work whenever I forgot.
It’s early, and I’d rather tell you where the edges are than sell around them. The workspace, the MCP connection and the render formats are live and in daily use.
If you’re putting a company brain in place, email me and I’ll help you set it up.
Where Tatara is today
Tatara is live, self-serve and early. Work started in April 2026 and the product ships continuously. The changelog lists what has landed and when.
A document in a brain opens as a page or, if it has a format, as a filterable base view, a typed table, a kanban board, a task tracker or a mind map. The file underneath stays plain Markdown either way, so nothing gets converted. Agents can also write self-contained HTML and build a brain’s homepage as a live dashboard.
- Name
- Tatara
- What it is
- A governed document store: the context layer where AI agents and people read a company’s knowledge.
- Founder
- Angus
- Work started
- April 2026
- Website
- www.tatara.app
- Documentation
- docs.tatara.app
- Agent protocol
- Model Context Protocol (MCP)
- Document format
- Markdown with YAML frontmatter
- Free plan
- 500 documents, no card required. See all plans.