Memory — your agent remembers you
Tell Agentic Hat something once and it sticks — or jot it down yourself. Your preferences carry across conversations, so you stop repeating yourself and every result lands closer to what you want.
How it works
You don’t switch memory on for it to work — it happens as you chat. When you express a lasting preference — a look, a default, a tone — the agent quietly saves it and reuses it the next time you start a conversation.
What it remembers
- Style & look preferences (e.g. warm, cinematic lighting)
- Defaults you keep asking for (e.g. vertical, ~5-second clips)
- Tone & wording you like the agent to use
- Recurring context about you or your brand
One-off details for a single task aren’t kept — only preferences that clearly apply again.
How memories are organized
Every memory is filed under a category and carries a status, a confidence score, and optional tags — so you can see at a glance what the agent knows and how sure it is.
- Preference — a style, look, or default you like
- Feedback — guidance you’ve given on how to work
- User — lasting facts about you or your brand
- Project — context about what you’re working on
Most memories stay Active. When a newer memory replaces an older one, the old one becomes Superseded; ones the agent retires show as Pruned.
View & manage your memory
Open Memory to search everything the agent has remembered, add your own, and remove anything you’d rather it forget.
You’re always in control. Memories aren’t edited in place — you add new ones or delete old ones, and every delete gives you a few seconds to undo. Remove anything you’d rather the agent forget.
Good to know
- Memories are created automatically as you chat — and you can add your own anytime.
- Deleting a memory has a brief undo window, so a mis-tap is easy to reverse.
- A Superseded status means a newer memory replaced an older one.
- Not sure why a result came out a certain way? Check Memory — a saved preference may be steering it.