How to give ChatGPT your company context
ChatGPT reads your company documents through Tatara over the Model Context Protocol, using ChatGPT’s developer mode. Developer mode is a full MCP client that can use every tool a server exposes, reads and writes alike. It is a switch you turn on rather than a connector you pick from a list, though, and OpenAI documents it as a web feature on paid accounts.
The address, whichever client you use: https://www.tatara.app/api/mcp
Does ChatGPT support MCP servers?
Yes, through developer mode, which is off by default. Developer mode gives ChatGPT on the web a full MCP client for servers you add yourself. It is not the same thing as the published apps in ChatGPT’s directory: those are connectors OpenAI has shipped, and Tatara is not one of them, so developer mode is the route.
How do you connect ChatGPT to Tatara?
You give the client Tatara’s address and approve the connection once. Tatara serves MCP over streamable HTTP at a single URL, and takes either a browser sign-in or an access token as the credential, so the steps are the same wherever you start.
- ROUTE 01
ChatGPT on the web
Turn on developer mode, then add Tatara as your own MCP app.
- 1.Turn on developer mode in your ChatGPT settings. On a Business or Enterprise workspace this is an admin setting rather than a personal one.
- 2.Create a developer-mode app pointing at https://www.tatara.app/api/mcp.
- 3.Sign in to Tatara and approve the consent screen naming the client.
Tatara serves MCP over streamable HTTP, which is one of the transports ChatGPT accepts, and supports Dynamic Client Registration, so there is no client ID and secret to generate and paste.
- ROUTE 02
The OpenAI API
Building on the API rather than using the ChatGPT app is a different, simpler path.
- 1.Pass https://www.tatara.app/api/mcp as a remote MCP tool on your request, with a Tatara access token as the bearer credential.
- 2.Create the token in Tatara under Settings, Access tokens. Only a workspace owner can issue one, and it is shown exactly once.
This route skips developer mode entirely. It is the right one for anything running unattended, because nothing has to open a browser.
The full procedure, including issuing and revoking access tokens, lives in the Tatara docs. For ChatGPT’s own screens, see OpenAI on developer mode and MCP apps.
What can ChatGPT do once it is connected?
It gets twenty-three tools against your brain, in five groups: finding and reading documents, writing them, organising them into folders, curating the tag and property vocabulary, and deleting or restoring things. The tools describe themselves, so a connected agent discovers what it needs when it needs it.
Every agent write is governed. A document needs a type, a title, a description and tags steered to your taxonomy, and a write that does not qualify comes back with what is wrong and how to fix it. The agent corrects itself inside its own loop instead of leaving a half-written document for someone to find later. Every save is stamped with the credential behind it, which is why an agent cannot pass itself off as a person.
Underneath, a brain is plain Markdown with YAML frontmatter: one file per concept, with links forming a graph, and nothing you could not open in an editor. You can also install the open source agent skills with npx skills add tatara-app/tatara-skills, which teach ChatGPT the workflow up front so its first write usually passes.
What are the limits with ChatGPT?
MCP support is not the same across assistants, so here is the honest version for this one.
- OpenAI documents developer mode as a web feature on paid ChatGPT accounts rather than free ones: Plus, Pro, Business, Enterprise and Education. Check OpenAI’s help page for the current list before you plan a rollout around it.
- OpenAI describes developer mode as intended for people comfortable configuring and testing apps, and warns about prompt injection and mistaken write actions. That caution is fair. On the write side, Tatara validates every agent write before it lands and stamps it with the credential behind it, so an agent’s writes are never mistaken for a person’s. It is still a switch to turn on deliberately.
- Tatara is not in ChatGPT’s published app directory, so there is no one-click install. The connection is one you add yourself.
- Every Tatara connection carries read and write access. There is no read-only credential.
ChatGPT and your company brain
Can ChatGPT access my company documents?
Yes, once you turn on developer mode and add the connection. Developer mode is ChatGPT’s full MCP client on the web, so it can use every tool Tatara exposes: searching your brain, reading a document with what links to it, and writing new ones back.
Is this the same as uploading files to a project or a custom GPT?
No. An upload puts a copy inside ChatGPT, and that copy is frozen the moment you make it. A connection leaves the document in Tatara, so ChatGPT reads whatever the current version says and your team edits it in one place. It also works the other way: ChatGPT can write back into the same brain your other agents read.
Do I need a paid ChatGPT account?
For the developer-mode route, yes. OpenAI lists it for Plus, Pro, Business, Enterprise and Education accounts rather than free ones. If you are calling the OpenAI API directly instead, developer mode does not come into it: you pass the MCP server as a tool with a Tatara access token.
Can ChatGPT write to my company brain?
Yes, and that is deliberate. Every agent write is checked before it lands: a document needs a type, a title, a description and tags steered to your taxonomy. Each save is stamped with the credential that made it, and locked documents refuse every agent write until a person unlocks them in the web app.
What does it cost to connect ChatGPT?
Nothing to start. The free plan holds 500 documents and does not ask for a card. Agents connect to it over MCP and can read your brain and make governed writes to it. Paid plans add headroom, and the pricing page has the current limits.