What is AI memory?
AI memory is information an AI system keeps about you and your work between conversations, such as preferences, facts and past decisions, and brings back when it is relevant.
Models do not remember on their own: each request starts empty. Memory is a store the system keeps beside the model and adds to the context when it matters.
How it works
The system decides what is worth keeping, because you asked or because a fact looked useful, and saves it with a note of where it came from. On a new request it looks up the stored items that relate to it and includes them, so the model can use them without you repeating yourself.
Examples
- Remembering that you prefer metric units.
- Knowing which of your computers holds your work files.
- Recalling a decision made in an earlier conversation about a project.
Limitations
- A wrong or outdated memory keeps shaping answers until it is removed.
- What is remembered is personal data, so you should be able to see, export and delete it.
- Memory adds to the context, so too much of it crowds out the request.
How Venta uses it
Venta shows what it has stored about you, with where each piece came from. You can download all of it, and switch memory off for a project.
Private details are removed before a memory is stored, and a memory can be limited to a project or set to expire.
Questions people ask first
- Can I see what Venta remembers about me?
- Yes. Memory lists what Venta has stored about you, with where each piece came from, and you can download it.
- Can I turn memory off?
- Yes, for a project: switch it off there, and nothing new is remembered from it.
- Is AI memory the same as a context window?
- No. The context window is what the model reads in one request. Memory is a separate store that is drawn into that window when it is relevant.