Memory
Memory is what lets an agent remember across conversations. Instead of starting fresh every time, an agent keeps notes — your preferences, an ongoing project's status, facts it's learned — and has them on hand in future chats.
How agents remember
An agent saves what matters as it works, organizing notes into topic files — one per subject, like "customer preferences" or "project status." It keeps an index of those topics automatically. There's no separate "memorizing" step that happens later; the agent captures things in the moment, during your conversation.
At the start of each session, the agent's memory comes back: while its notes stay compact they're loaded in full, and once they outgrow that budget it gets the index instead and reads any topic on demand. Either way, it walks into every conversation already knowing what it learned before.
Shared and private memory
Memory comes in two scopes:
- Shared (agent) memory — visible to everyone who works with the agent, and something the team builds up together. Good for facts about a shared project or process.
- Private (user) memory — your own notes with an agent, visible only to you.
Which scopes an agent uses depends on its mode (see Agents): a personal-only assistant keeps just your private memory, a shared agent keeps team memory, and most collaborative agents keep both.
A gentle nudge to remember
If an agent has been chatting for a while without saving anything, OtoDock quietly reminds it to capture what's worth keeping. The reminder is silent — you never see it.
Managing memory
- Clear all your private memory — one button in your settings wipes your private memory across every agent at once.
- Managers can clear an agent's shared memory from the agent's settings.
- When you're removed from an agent, your private memory for it is cleared too.
Read and write access follows the same roles as everything else: viewers can read shared memory, editors and managers can contribute to it, and no one can see another person's private notes.