​
Memory Storage

To persist memories across sessions and execution cycles, store memories in a persistent storage like a database.

If you’re using Memory in production, persistent storage is critical as you’d want to retain user memories across application restarts.

Agno’s memory system supports multiple persistent storage options.

​
Storage Options

The Memory class supports different backend storage options through a pluggable database interface. Currently, Agno provides:

  1. SQLite Storage
  2. PostgreSQL Storage
  3. MongoDB Storage
  4. Redis Storage

​
Setting Up Storage

To configure memory storage, you’ll need to create a database instance and pass it to the Memory constructor:

from agno.memory.v2.memory import Memory
from agno.memory.v2.db.sqlite import SqliteMemoryDb

# Create a SQLite database for memory
memory_db = SqliteMemoryDb(
    table_name="memories",  # The table name to use
    db_file="path/to/memory.db"  # The SQLite database file
)

# Initialize Memory with the storage backend
memory = Memory(db=memory_db)

​
Data Model

When using persistent storage, the Memory system stores:

  • User Memories - Facts and insights about users
  • Last Updated Timestamps - To track when memories were last modified
  • Memory IDs - Unique identifiers for each memory

​
Storage Examples

sqlite_memory.py
from agno.memory.v2.memory import Memory
from agno.memory.v2.db.sqlite import SqliteMemoryDb
from agno.memory.v2.schema import UserMemory

# Create a SQLite memory database
memory_db = SqliteMemoryDb(
    table_name="user_memories",
    db_file="tmp/memory.db"
)

# Initialize Memory with the storage backend
memory = Memory(db=memory_db)

# Add a user memory that will persist across restarts
user_id = "user@example.com"
memory.add_user_memory(
    memory=UserMemory(
        memory="The user prefers dark mode in applications",
        topics=["preferences", "ui"]
    ),
    user_id=user_id
)

# Retrieve memories (these will be loaded from the database)
user_memories = memory.get_user_memories(user_id=user_id)
for m in user_memories:
    print(f"Memory: {m.memory}")
    print(f"Topics: {m.topics}")
    print(f"Last Updated: {m.last_updated}")
postgres_memory.py
from agno.memory.v2.memory import Memory
from agno.memory.v2.db.postgres import PostgresMemoryDb
from agno.memory.v2.schema import UserMemory

# Create a PostgreSQL memory database
memory_db = PostgresMemoryDb(
    table_name="user_memories",
    connection_string="postgresql://user:password@localhost:5432/mydb"
)

# Initialize Memory with the storage backend
memory = Memory(db=memory_db)

# Add user memories
user_id = "user@example.com"
memory.add_user_memory(
    memory=UserMemory(
        memory="The user has a premium subscription",
        topics=["subscription", "account"]
    ),
    user_id=user_id
)

# Memory operations work the same regardless of the backend
print(f"User has {len(memory.get_user_memories(user_id=user_id))} memories stored")

​
Integrating with Agent Storage

When building agents with memory, you’ll often want to store both agent sessions and memories. Agno makes this easy by allowing you to configure both storage systems:

from agno.agent import Agent
from agno.memory.v2.memory import Memory
from agno.memory.v2.db.sqlite import SqliteMemoryDb
from agno.models.openai import OpenAIChat
from agno.storage.sqlite import SqliteStorage

# Create memory storage
memory_db = SqliteMemoryDb(
    table_name="memories",
    db_file="tmp/memory.db"
)
memory = Memory(db=memory_db)

# Create agent storage
agent_storage = SqliteStorage(
    table_name="agent_sessions",
    db_file="tmp/agent_storage.db"
)

# Create agent with both memory and storage
agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    memory=memory,
    storage=agent_storage,
    enable_user_memories=True,
)

​
Memory Management

When using persistent storage, the Memory system offers several functions to manage stored memories:

# Delete a specific memory
memory.delete_user_memory(user_id="user@example.com", memory_id="memory_123")

# Replace/update a memory
memory.replace_user_memory(
    memory_id="memory_123",
    memory=UserMemory(memory="Updated information about the user"),
    user_id="user@example.com"
)

# Clear all memories
memory.clear()

​
Developer Resources

  • Find reference documentation for memory storage here