Module 5: Agent Memory¶

Give your agent memory. Session context tracks the current conversation. Long term preferences persist across conversations.

How it works¶

Memory type	What it stores	Lifetime
Session	Every message in the current conversation	Expires after the session ends
Long term	User preferences and facts	Persists across conversations

Cloud setup¶

1. Create an Agent Memory store¶

In the Redis Cloud console, find Agent Memory in the left sidebar.
Click Quick Create.

2. Copy your API key¶

Save this — you won't see it again.

Agent Memory service key

3. Get your Endpoint and Store ID¶

After dismissing the key dialog, find both on the store details page:

Agent Memory credentials

4. Configure environment¶

Add to your .env:

MEMORY_API_BASE_URL=<endpoint from step 3>
MEMORY_STORE_ID=<store-id from step 3>
MEMORY_API_KEY=<API key from step 2>

5. Seed long term memories¶

macOS / LinuxWindows (PowerShell)

make seed-memories

.\workshop.ps1 seed-memories

This seeds two facts for the demo customer: "Prefers contactless delivery" and "Likes spicy food."

Exercise¶

Open exercises/digital-native/agent_memory.py

`long_term_search_payload(text, owner_id, session_id, limit)`¶

This tells Agent Memory how to search for stored user preferences. When a conversation starts, the agent calls this to recall what it knows about the user.

def long_term_search_payload(self, *, text, owner_id, session_id, limit):
    return {
        "text": text,                                        # what to search for
        "similarityThreshold": 0.2,                          # how close a match must be
        "filterOp": "all",                                   # all filters must match
        "limit": limit or 5,                                 # max memories to return
        "filter": {
            "ownerId": {"eq": sanitize_owner_id(owner_id)},  # scope to this user
            "namespace": {"eq": "digital-native-demo"},              # scope to this app
        },
    }

Fill in the method body with the dict above.

filter scopes results so each user only sees their own memories
sanitize_owner_id is already imported at the top of the file

Try it¶

Restart with make dev, then open localhost:3040.

Ask: "What should I order?" — the agent recalls "contactless delivery" and "spicy food"
Ask a follow up: "From what restaurant?" — the agent remembers what you just discussed

Verify¶

Agent recalls seeded preferences ("contactless delivery" and "spicy food")
Multi turn conversations maintain context
Activity panel shows session and long term memory lookups