Use casesUse case
A support knowledge base your team and AI tools both use
A support team drops its docs and past answers into an engine's Files, saves new answers and decisions as memories as they happen, and then both people (with Ask) and AI tools (over MCP) answer from the same knowledge, with the source of each answer named.
What it does
Support answers are scattered over a help centre, old tickets and senior agents’ memory. In one engine they become something every agent, and every agent’s AI tool, can search by meaning. When someone works out a new answer, their AI tool saves it as a note or a decision, so the next person finds it.
Set it up
- Make an engine for support knowledge, with a purpose that says what belongs: product behaviour, known issues, policies and how they were decided.
- Drop the help centre export, manuals and policy documents into Files. A folder name becomes a tag, so you can find the set again.
- Connect each agent’s AI tool (ChatGPT, Claude, Cursor) at Can read, or Can read and save for the people who maintain answers.
- Invite the support team. People are free on every plan, and a viewer’s connections only read.
Using it
In the app, Ask a question and point at any number in the answer to see which memory it came from and who saved it. In an AI tool, the same question searches every engine the connection reaches and every result names its engine. Every change records who made it, with which tool, and what it was before.
Questions
Will the answers make things up?
Ask answers only from the engine's memories and names the ones it used. When the answer is not there it says so; it does not look anything up on the web.
Which files can we import?
Files takes files, whole folders or a .zip. Text and PDFs become searchable and pictures are described once. Each file shows whether it is searchable, and why not when it is not.