AI Concept Atlas
FIELD NOTES / 01
Guided learning · 6 lessons

What the model can see

Distinguish current context, retrieved evidence, and retained memory.

Lesson 6 of 6

Memory

Evidence for one answer differs from information retained for later use.

In agent applications, memory is retained information that can be reused later. It can include conversation state within a session or information saved across sessions, depending on the application.

[LangChain]

In practice

A tutoring app saves a learner’s preferred explanation style and loads that preference when a later lesson starts.

[LangChain]

A little deeper

Frameworks distinguish short-term conversation state from longer-term stored information. Retaining data and selecting what enters the next model input are separate responsibilities. [LangChain]

A common mix-up

Memory means the model sees all previous conversations.

The application decides what to store and what to recall. [LangChain]

Helpful to know: Context Window

Explore Memory and its connections ↗

← Previous lesson