Guide
What AI apps actually remember about you
And the parts you are not shown.
Most AI products now keep some record of you. Very few show you all of it. Here is what is typically stored, and how to check.
Three different things get called memory
Conversation history
The transcript of what you said. Useful as an archive, but it only helps if you go back and search it. It is storage, not recall.
Notes the product decided to keep
Facts extracted from your conversations: your job, your city, that you prefer short answers. This is the part usually meant by "memory", and it is the part most often partially hidden. Some products show a list, some show a summary, some show nothing.
Account and usage data
Not memory at all in the useful sense, but it is data about you, and it is worth knowing whether it feeds anything else, such as advertising.
The questions that actually matter
- Can I see all of it? Not a summary. The list, line by line.
- Can I correct it? Wrong facts are worse than no facts, because they quietly shape every answer.
- Can I delete it, for real? One fact, all of it, and the account, with no hidden archive.
- Does it train the model? There is a large difference between using your words to answer you and using them to build a product.
- Can I take it with me? If it cannot be exported, it is not yours.
- Who profits from knowing me? A subscription business and an advertising business want different amounts of information about you.
Try this today. Open the AI app you use most and look for a memory or personalisation setting. Read what it lists. The gap between what it knows about you and what it is willing to show you is the thing worth paying attention to.
Why memory locked in one product is a problem
Even when it is accurate and well handled, memory held inside a single AI app only works in that app. Change tools and you are a stranger again, which quietly makes switching expensive. That is convenient for the company holding the notes.
Memory kept on your side, as a layer rather than a feature, avoids that. Same record, whichever model you use, and it leaves with you.
How Tessera answers those six questions
Everything is visible line by line, editable and genuinely deletable, with a full copy of your data available on request. Your conversations reach the AI models on business tiers where conversations are not used to train the models by default. There is no ads business. The detail, in plain language.
Questions
Related questions
What do AI apps remember about me?
Usually three things: the transcript of your conversations, facts extracted from them such as your job or preferences, and account and usage data. The middle category is what most products mean by memory, and it is the one least often shown to you in full.
Can I see everything an AI remembers about me?
It depends on the product. Some show a full list, some a summary, some nothing at all. If you cannot read it line by line, you cannot correct what it got wrong.
Does AI memory train the model?
In some products it can. There is a real difference between using your words to answer you and using them to improve a model. Check the terms, not the marketing.
Stop explaining yourself to AI.
Ten days free, no card required. iPhone and web.