Create a workspace
Open Manex Team Brain, create or restore a Team Brain workspace, and import the documents or memory you want available.
Manex Team Brain can expose a secure MCP connector so Claude or ChatGPT can answer from your workspace memory, cite compact evidence, and save useful corrections or decisions back into Manex.
The assistant calls Manex Brain through MCP, retrieves the relevant saved decisions, corrections, memories, and document chunks, then writes the final answer in its own style.
If a domain expert corrects an answer in Claude or ChatGPT, the connector can save that correction to Manex so future answers prefer the corrected context.
The connector is generated from the Team app and scoped to one workspace. Use the URL Manex gives you, not a generic server URL.
Open Manex Team Brain, create or restore a Team Brain workspace, and import the documents or memory you want available.
In Team Share, generate the Claude/ChatGPT connector URL. Manex creates a workspace-scoped token for that connector.
Paste the generated URL into Claude custom connectors or ChatGPT MCP settings, then enable the Manex Brain tools.
Prompt the assistant to use Manex Brain when the answer should come from your team’s saved evidence and memory.
The connector is designed around Manex as the memory layer and Claude or ChatGPT as the synthesis layer.
Retrieve grounded answers from the workspace’s synced memories, decisions, corrections, and selected document chunks.
Fetch relevant evidence snippets so the assistant can cite where the answer came from and mention uncertainty.
Store useful conversation outcomes, expert corrections, or final team decisions back into Manex Brain.
These prompts help Claude or ChatGPT route the question through Manex Brain instead of answering from general model memory.
Each generated connector URL acts like a password for one Team Brain workspace. Different workspace leaders generate different connector URLs and do not share memory across workspaces.
A connector generated by one team leader only retrieves context from that leader’s workspace.
Workspace owners can rotate the connector URL or revoke it when access should stop.
Manex returns compact evidence packets, not huge unfiltered JSON or casual copies of every document.
No. Manex remains the grounding layer. Claude or ChatGPT is the synthesis layer that writes the final answer using Manex context.
Manex automatically syncs MCP context after workspace restore, document indexing, post-answer learning, and memory updates. Manual sync is still available for force refresh.
No. The connector token is scoped to a workspace. A different leader and workspace generates a different connector URL and context boundary.
Start in Manex Team Brain, generate the connector URL, then ask Claude or ChatGPT to use Manex Brain for grounded answers.