Conversation Memory#

Overview#

You can persist and load memory by using Conversation Memory Drivers. You can build drivers for your own data stores by extending BaseConversationMemoryDriver .

Conversation Memory Drivers#

Gen AI Builder#

The GriptapeCloudConversationMemoryDriver allows you to persist Conversation Memory in Gen AI Builder. It provides seamless integration with Gen AI Builder’s cloud-based Threads and Messages resources.

import os

from griptape.drivers.memory.conversation.griptape_cloud import GriptapeCloudConversationMemoryDriver
from griptape.memory.structure import ConversationMemory
from griptape.structures import Agent

cloud_conversation_driver = GriptapeCloudConversationMemoryDriver(
    api_key=os.environ["GT_CLOUD_API_KEY"],
    alias="my_thread_alias",
)
agent = Agent(conversation_memory=ConversationMemory(conversation_memory_driver=cloud_conversation_driver))

agent.run("My name is Jeff.")
agent.run("What is my name?")

Local#

The LocalConversationMemoryDriver allows you to persist Conversation Memory in a local JSON file.

from griptape.drivers.memory.conversation.local import LocalConversationMemoryDriver
from griptape.memory.structure import ConversationMemory
from griptape.structures import Agent

local_driver = LocalConversationMemoryDriver(persist_file="memory.json")
agent = Agent(conversation_memory=ConversationMemory(conversation_memory_driver=local_driver))

agent.run("Surfing is my favorite sport.")
agent.run("What is my favorite sport?")

Amazon DynamoDb#

The AmazonDynamoDbConversationMemoryDriver allows you to persist Conversation Memory in Amazon DynamoDb .

import os
import uuid

from griptape.drivers.memory.conversation.amazon_dynamodb import AmazonDynamoDbConversationMemoryDriver
from griptape.memory.structure import ConversationMemory
from griptape.structures import Agent

conversation_id = uuid.uuid4().hex
dynamodb_driver = AmazonDynamoDbConversationMemoryDriver(
    table_name=os.environ["DYNAMODB_TABLE_NAME"],
    partition_key="id",
    value_attribute_key="memory",
    partition_key_value=conversation_id,
)

agent = Agent(conversation_memory=ConversationMemory(conversation_memory_driver=dynamodb_driver))

agent.run("My name is Jeff.")
agent.run("What is my name?")

Optional parameters sort_key and sort_key_value can be supplied for tables with a composite primary key.

Redis#

The RedisConversationMemoryDriver allows you to persist Conversation Memory in Redis .

import os
import uuid

from griptape.drivers.memory.conversation.redis import RedisConversationMemoryDriver
from griptape.memory.structure import ConversationMemory
from griptape.structures import Agent

conversation_id = uuid.uuid4().hex
redis_conversation_driver = RedisConversationMemoryDriver(
    host=os.environ["REDIS_HOST"],
    port=int(os.environ["REDIS_PORT"]),
    username=os.environ["REDIS_USERNAME"],
    password=os.environ["REDIS_PASSWORD"],
    index=os.environ["REDIS_INDEX"],
    conversation_id=conversation_id,
)

agent = Agent(conversation_memory=ConversationMemory(conversation_memory_driver=redis_conversation_driver))

agent.run("My name is Jeff.")
agent.run("What is my name?")

Conversation Memory

Contents