What is AI memory?

By the Variask team · Updated

Definition

AI memory is when an AI assistant remembers facts from past conversations, such as a user's location, job or preferences, and uses them in later answers. It means two people can get different recommendations for the same question.

Also called Personalised AI answers

How it works

  • Assistants can store facts the user shares or allows.
  • Later answers can take those facts into account.
  • Users can usually view, edit or turn off memory.
  • Tracking tools ask without memory, as a fresh user would.

What it means for brands

  • Personalisation makes single checks even less reliable.
  • Measure as a neutral user in each market for a fair baseline.

Questions

AI memory:
common questions

It can, for example by favouring local options or past choices. A fresh baseline shows what new buyers see.

Not for real users. Tracking tools measure a clean baseline by country, city and language.

Related terms

Keep exploring

Engines and surfaces

Conversational search

Conversational search is searching by asking questions in natural language and following up, as in a chat, instead of typing short keywords.

Read

Measurement

Answer volatility

Answer volatility is how much AI answers to the same question change from one run to the next, or from week to week. High volatility means a single check can be misleading, so scores need many answers and a likely range.

Read

How AI search works

System prompt

A system prompt is the hidden instruction an AI product gives its model before the user's question, setting its role, tone and rules. It shapes how answers are written, such as whether to name brands or cite sources, but users do not see it.

Read
Full glossary