Core concepts

Eleven ideas carry the whole product. Learn these and the rest reads easily.

Workspace#

A workspace is your organization's own, isolated area of Breno. Everything you put in belongs to one workspace, and nothing in it is visible to another. People join a workspace with a role (Owner, Admin, Member or Viewer; see Roles and permissions). The database itself enforces the isolation, not only the application: a query that forgot to scope itself to a workspace still cannot return another workspace's rows.

Item#

An item is one piece of knowledge: an uploaded file, a note, a saved link, a meeting, or something synced from a connected tool. Every item keeps:

• an owner (who added it) and the dates it was created and changed;
• its provenance: where it came from, the author, a link to the original, and when it last changed at the source;
• a visibility (the whole workspace, or Only me) and a sensitivity label;
• an optional space, tags and collections;
• a version history, for notes and for synced items that change at the source.

See Organize knowledge.

Source#

A source is a connected tool that Breno reads from, such as Slack or Google Drive. Items that arrive from a source stay linked to it, so an edit there updates the same item and adds a version, and a deletion there moves it to trash. See Provenance and sync.

Space#

A space is a group of items with its own members. Putting an item in a space limits who can read it to the people in that space (and anyone cleared for it by role). Spaces are one of the two controls on who can see what; the other is sensitivity.

Sensitivity#

A sensitivity label says how careful to be with an item: public, internal, confidential or restricted. A person only sees an item if their clearance allows its label. See How access works.

The read rule#

Every way of reading knowledge (search, answers, the graph, exports, MCP, the API) goes through one shared read rule that applies spaces, sensitivity labels and private items. It is the same code everywhere, and no language model takes part in deciding access.

Memory#

Memory is what Breno understands from your items: decisions (who decided what, when and why), entities and the relationships between them, and a timeline. Breno proposes connections and possible duplicates; nothing changes in the graph until you approve it. See Memory and decisions and Memory graph and time travel.

Citation#

A citation is a numbered pointer from an answer to the passage it was built from, carrying the source, author and date and a link to the original. See How answers are built.

Token#

A personal token (it starts with sirmx_mcp_) lets your own tools use Breno as you, in one workspace, with the scopes you chose. The same token works on the REST API and the MCP server. See Tokens.

MCP#

The Model Context Protocol is an open standard for giving AI tools access to data and actions. Breno runs an MCP server, so a client that speaks MCP can search, read, ask, export and capture into your workspace. See MCP server.

Coins#

Breno is paid for with coins, not a subscription. Reading, searching and viewing are free; asking a question and adding documents spend coins. See Billing and coins.