How memory works
What Esment keeps, how it gets there, and how your assistants use it in every new chat.
What a memory is
A memory is one small thing worth remembering: a fact, a preference, a decision or a piece of context. “I prefer short replies”, “my company is called Acme”, “we chose blue for the logo”. Each one is short, and together they build a picture your assistants can draw on.
Every memory also carries a little information about itself: when it was saved, what kind it is (a Fact, Knowledge, Moment or How-to), its tags, and which assistant saved it. You see this on each memory in the list.


How memories get saved
There are three ways, and you can mix them freely.
- Automatically. While you chat with a connected assistant, it saves durable details it notices, such as your role, a preference or a decision. You do not have to do anything.
- Because you asked. Say “remember that…” and the assistant saves it right away. You can also ask “forget that” to remove something.
- By you. Add a memory yourself with Add memory on the web or New Memory in the Mac app. On the Mac, ⌥C opens a quick note from anywhere.
Remember that I prefer short replies and plain language.
Voice notes and imported conversations end up as memories too. See Voice notes and Capture & import.
Recall at the start of every chat
An assistant that has to decide for itself when to look something up often does not. So Esment prepares a short session brief in advance: the people, projects and preferences that matter to you, your pinned memories, your Memory Blocks and the work you did recently.
A connected assistant reads that brief before it answers, so a new chat starts from what you already know rather than from nothing. Later in the chat it can also search your memory for a specific fact.
On the Mac, search and recall run on your computer. Nothing you write is sent anywhere to find the right memory, and it works in English, Spanish, German, French and Italian, even when you saved something in one language and ask in another.
Which assistant saved it
When an assistant saves a memory, Esment records which one it was and shows it next to the memory, for example via ChatGPT or via Claude. If a memory looks wrong, this tells you where to look.
Memories you add yourself, or that were saved before this feature existed, do not show a name. Anything you see can be edited or forgotten.
Auto summaries
Besides the memories you save, Esment writes a short summary for the people, projects and topics that come up most. These are labelled Auto summary, so you can tell what Esment wrote from what you or your assistants saved.
- Each person, project or topic keeps one current summary. Older versions do not pile up.
- A summary is rewritten only when there is real news to add.
- You can read them like any other memory. They feed the Library too.
The life of a memory
Not every memory should live forever. Each one has a state, shown as a label on the memory, so your memory stays fresh instead of piling up.
- New: just saved.
- Active: in regular use and available to your assistants.
- Merged: combined with a related memory to avoid duplicates.
- Archived: set aside as it becomes less relevant, but still kept. You can bring it back with Restore.
- Forgotten: no longer offered to assistants, either because you forgot it or because it aged out.
Most of this happens on its own, so you never have to tidy up. A pinned memory is the exception: it is never archived automatically. The Mac app calls this Evergreen.
Where to go next
- Memory Blocks: always-on notes about you and your projects.
- History & Time Travel: see what your assistants learned, day by day.
- Knowledge graph: the people and projects in your memory.
- Voice notes: say it once instead of typing.