How do I stop repeating myself to AI agents?
Write it down once, somewhere every agent checks. With Clyde, you say it once, the agent saves a short note, and every connected agent recalls it when it's relevant. Two weeks and three apps later, nobody asks you again.
Why it keeps happening
Agents don't carry anything between chats, between apps or between computers. So you're the memory. You end up explaining the same naming rule, the same deploy step and the same "no, the other database" a thousand times.
Say it once
- Tell the agent: "remember that Bob only likes sage green." It saves a note and gives you the ID.
- Any agent working on Bob's stuff recalls it, because it matches what it's doing.
- If it changes, tell an agent and the new note replaces the old one.
Recall hands back a small packet of matching notes (up to 16 settled, plus up to 3 on trial), so your agent doesn't drown in everything you ever said.
Next
See saving and recalling notes, or start at the docs.
Questions
What should I save?
Decisions, preferences, how things are done, and why something broke. Short notes, under 500 characters, with longer stuff attached as files.
Do I have to ask the agent to save things?
No. Agents save what they figure out as they go. You can also tell one to remember something.
Can it remember too much?
New notes start on trial for 3 recalls. Ones that don't help fade, and they fade by how often you recall, not by the calendar.