Agentic SystemsProAdvanced

Long-Term Memory for AI Agents

Persist agent knowledge beyond the context window — vector stores, key-value caches, and structured stores for cross-session recall.

By Amit UpadhyayPublished 18 MayUpdated 13 Oct 202612 minArticle 4 of 12 in Agentic Systems

On this page

The idea at a glance

The problem

When

An assistant should remember stable facts about each user, such as preferences, defaults and constraints, across sessions

The obvious move

Keeping every transcript forever and retrieving from all of it

Fails because

Contradictory old statements resurface, stale preferences override new ones, and you cannot honour a deletion request without searching everything

The fix

Long-term memory works by

A small, explicit set of facts per user, written back when the user states or confirms them, updated rather than appended, and deletable on request

At the cost of

A write-back step, rules for conflicting facts, and deletion that has to reach every copy

The scenario every mode follows

““Send it to my office like last time.””

The shopping and support assistant, used by repeat customers
~200,000 returning customers
claude-sonnet-5-5

Where the obvious approach breaks

A 1-month deadline for erasure requests (GDPR Art. 12(3))

If you keep all transcripts and retrieve whatever matches.

Under the GDPR, a request to erase personal data must be answered without undue delay and at the latest within one month. With memory spread across raw transcripts, embeddings and caches, nobody can say with confidence that a user has been fully forgotten.

Helps to know firstEpisodic memory

What an Assistant Should Remember, and Forget

Long-term memory is what the agent remembers across conversations. Without it, every session starts from zero — the user re-explains their job, their tone preferences, the project context, and the bug they reported yesterday. With it, the agent feels like a coworker; without it, it feels like a customer-service hold queue.

Before you store everything

Users complain the agent "forgets everything" between sessions
You stuffed the whole user history into the system prompt and it costs $0.20 per question
You don't know how to forget when the user asks you to

Pro deep dive12 min full read

Keep reading with Pro

This is one of our advanced production deep dives. Pro unlocks the rest of it, every other deep dive, and the See, Animate, Test and Build modes on every topic.

What's inside