Teach it once: one memory for the whole team

Here is the test for whether an AI is a teammate or a chatbot: correct it, come back tomorrow, and see if the correction is still there. With most tools it is not. Every task starts from zero and every rule has to be repeated, by every person, every time.

Cedric keeps one memory for the workspace. Tell it something once and it stays known: to you, to your teammates, and to every task that comes later. The tenth task is faster and better informed than the first, and what it learns helping one person is already known when a colleague asks the next day.

How it works

Three kinds of things live in the memory. Facts you tell it, like the one above. Facts it quietly keeps from the work it does, so nobody has to say "the invoice number is on the second line" twice. And instructions, which sit above memories: workspace rules and personal preferences that shape how Cedric works, not just what it knows.

Every memory is either shared with the workspace or visible only to you, and nothing decides that silently. Private conversations stay private. Anything learned through a private connection, your own inbox for instance, stays yours.

All of it is listed on one page, the Brain, one line per memory with where it came from. Click to read it, one click to delete it. Delete means delete.

How to use it on purpose

Start with the three things your team re-explains most. The naming rule for customers. Where the numbers live. How the weekly note should read. Say each one once, in the channel, and watch the next draft follow it.

Then bring the documents. Drop in the onboarding doc or the style guide and ask Cedric to learn it. If you taught another assistant about the company already, paste what it knows; it carries over.