All use cases

Memory by use case

AI memory for customer support agents

Stop asking customers to explain the same problem a third time.

Why memory matters here

Support is the place people notice forgetting most. They have already said which plan they are on, what they tried and what did not work. An agent that keeps those facts between chats can start from the last step instead of the first, and a human who takes over reads a short summary instead of a long transcript.

Who a memory belongs to

One memory per customer, with a second, shared memory per account when several people from the same company write in.

The moment it shows

The customer comes back two days later and the agent opens with the fix that was still pending, not with "How can I help?".

What to remember

  • Product, plan and setup details the customer has already given.
  • What was tried, what failed and what was promised.
  • How the customer likes to be contacted and in which language.

What to forget

  • Card numbers, passwords and one-time codes: never store them, even if the customer types them.
  • Frustration from one bad day. Keep the issue, drop the mood.
  • Workarounds for bugs that have since been fixed.

See recall before you build it

Add a few facts from your own use case and ask a question.

Customer support agents: common questions

Is a ticket history the same as agent memory?

No. A ticket history is a full record for people to read. Agent memory is a short set of facts taken from that record, small enough to put in a prompt on every turn.

Should a support agent remember across different customers?

Only general knowledge, such as a known bug and its fix. Anything a customer said about themselves stays in that customer's memory.

Other use cases

Pack the doko. Ask it anything.

Nine memories, one question, no account. See which facts an agent would carry into its next answer.