Skip to main content

Zep Cloud Memory

Recall, understand, and extract data from chat histories. Power personalized AI experiences.

Zep is a long-term memory service for AI Assistant apps. With Zep, you can provide AI assistants with the ability to recall past conversations, no matter how distant, while also reducing hallucinations, latency, and cost.

See Zep Cloud Installation Guide and more Zep Cloud Langchain Examples

Example​

This notebook demonstrates how to use Zep as memory for your chatbot.

We'll demonstrate:

  1. Adding conversation history to Zep.
  2. Running an agent and having message automatically added to the store.
  3. Viewing the enriched messages.
  4. Vector search over the conversation history.
from uuid import uuid4

from langchain.agents import AgentType, Tool, initialize_agent
from langchain_community.memory.zep_cloud_memory import ZepCloudMemory
from langchain_community.retrievers import ZepCloudRetriever
from langchain_community.utilities import WikipediaAPIWrapper
from langchain_core.messages import AIMessage, HumanMessage
from langchain_openai import OpenAI

session_id = str(uuid4()) # This is a unique identifier for the session
---------------------------------------------------------------------------
``````output
AttributeError Traceback (most recent call last)
``````output
Cell In[3], line 8
6 from langchain_community.utilities import WikipediaAPIWrapper
7 from langchain_core.messages import AIMessage, HumanMessage
----> 8 from langchain_openai import OpenAI
10 session_id = str(uuid4()) # This is a unique identifier for the session
``````output
File ~/job/integrations/langchain/libs/partners/openai/langchain_openai/__init__.py:1
----> 1 from langchain_openai.chat_models import (
2 AzureChatOpenAI,
3 ChatOpenAI,
4 )
5 from langchain_openai.embeddings import (
6 AzureOpenAIEmbeddings,
7 OpenAIEmbeddings,
8 )
9 from langchain_openai.llms import AzureOpenAI, OpenAI
``````output
File ~/job/integrations/langchain/libs/partners/openai/langchain_openai/chat_models/__init__.py:1
----> 1 from langchain_openai.chat_models.azure import AzureChatOpenAI
2 from langchain_openai.chat_models.base import ChatOpenAI
4 __all__ = [
5 "ChatOpenAI",
6 "AzureChatOpenAI",
7 ]
``````output
File ~/job/integrations/langchain/libs/partners/openai/langchain_openai/chat_models/azure.py:8
5 import os
6 from typing import Any, Callable, Dict, List, Optional, Union
----> 8 import openai
9 from langchain_core.outputs import ChatResult
10 from langchain_core.pydantic_v1 import Field, SecretStr, root_validator
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/openai/__init__.py:8
5 import os as _os
6 from typing_extensions import override
----> 8 from . import types
9 from ._types import NOT_GIVEN, NoneType, NotGiven, Transport, ProxiesTypes
10 from ._utils import file_from_path
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/openai/types/__init__.py:5
1 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
3 from __future__ import annotations
----> 5 from .batch import Batch as Batch
6 from .image import Image as Image
7 from .model import Model as Model
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/openai/types/batch.py:7
4 from typing import List, Optional
5 from typing_extensions import Literal
----> 7 from .._models import BaseModel
8 from .batch_error import BatchError
9 from .batch_request_counts import BatchRequestCounts
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/openai/_models.py:667
662 json_data: Body
663 extra_json: AnyMapping
666 @final
--> 667 class FinalRequestOptions(pydantic.BaseModel):
668 method: str
669 url: str
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_model_construction.py:202, in __new__(mcs, cls_name, bases, namespace, __pydantic_generic_metadata__, __pydantic_reset_parent_namespace__, _create_model_module, **kwargs)
199 super(cls, cls).__pydantic_init_subclass__(**kwargs) # type: ignore[misc]
200 return cls
201 else:
--> 202 # this is the BaseModel class itself being created, no logic required
203 return super().__new__(mcs, cls_name, bases, namespace, **kwargs)
205 if not typing.TYPE_CHECKING: # pragma: no branch
206 # We put `__getattr__` in a non-TYPE_CHECKING block because otherwise, mypy allows arbitrary attribute access
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_model_construction.py:539, in complete_model_class(cls, cls_name, config_wrapper, raise_errors, types_namespace, create_model_module)
532 # debug(schema)
533 cls.__pydantic_core_schema__ = schema
535 cls.__pydantic_validator__ = create_schema_validator(
536 schema,
537 cls,
538 create_model_module or cls.__module__,
--> 539 cls.__qualname__,
540 'create_model' if create_model_module else 'BaseModel',
541 core_config,
542 config_wrapper.plugin_settings,
543 )
544 cls.__pydantic_serializer__ = SchemaSerializer(schema, core_config)
545 cls.__pydantic_complete__ = True
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/main.py:626, in __get_pydantic_core_schema__(cls, source, handler)
611 @classmethod
612 def __pydantic_init_subclass__(cls, **kwargs: Any) -> None:
613 """This is intended to behave just like `__init_subclass__`, but is called by `ModelMetaclass`
614 only after the class is actually fully initialized. In particular, attributes like `model_fields` will
615 be present when this is called.
616
617 This is necessary because `__init_subclass__` will always be called by `type.__new__`,
618 and it would require a prohibitively large refactor to the `ModelMetaclass` to ensure that
619 `type.__new__` was called in such a manner that the class would already be sufficiently initialized.
620
621 This will receive the same `kwargs` that would be passed to the standard `__init_subclass__`, namely,
622 any kwargs passed to the class definition that aren't used internally by pydantic.
623
624 Args:
625 **kwargs: Any keyword arguments passed to the class definition that aren't used internally
--> 626 by pydantic.
627 """
628 pass
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_schema_generation_shared.py:82, in CallbackGetCoreSchemaHandler.__call__(self, source_type)
81 def __call__(self, __source_type: Any) -> core_schema.CoreSchema:
---> 82 schema = self._handler(__source_type)
83 ref = schema.get('ref')
84 if self._ref_mode == 'to-def':
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:502, in generate_schema(self, obj, from_dunder_get_core_schema)
498 schema = _add_custom_serialization_from_json_encoders(self._config_wrapper.json_encoders, obj, schema)
500 schema = self._post_process_generated_schema(schema)
--> 502 return schema
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:753, in _generate_schema_inner(self, obj)
749 def match_type(self, obj: Any) -> core_schema.CoreSchema: # noqa: C901
750 """Main mapping of types to schemas.
751
752 The general structure is a series of if statements starting with the simple cases
--> 753 (non-generic primitive types) and then handling generics and other more complex cases.
754
755 Each case either generates a schema directly, calls into a public user-overridable method
756 (like `GenerateSchema.tuple_variable_schema`) or calls into a private method that handles some
757 boilerplate before calling into the user-facing method (e.g. `GenerateSchema._tuple_schema`).
758
759 The idea is that we'll evolve this into adding more and more user facing methods over time
760 as they get requested and we figure out what the right API for them is.
761 """
762 if obj is str:
763 return self.str_schema()
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:580, in _model_schema(self, cls)
574 inner_schema = new_inner_schema
575 inner_schema = apply_model_validators(inner_schema, model_validators, 'inner')
577 model_schema = core_schema.model_schema(
578 cls,
579 inner_schema,
--> 580 custom_init=getattr(cls, '__pydantic_custom_init__', None),
581 root_model=False,
582 post_init=getattr(cls, '__pydantic_post_init__', None),
583 config=core_config,
584 ref=model_ref,
585 metadata=metadata,
586 )
588 schema = self._apply_model_serializers(model_schema, decorators.model_serializers.values())
589 schema = apply_model_validators(schema, model_validators, 'outer')
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:580, in <dictcomp>(.0)
574 inner_schema = new_inner_schema
575 inner_schema = apply_model_validators(inner_schema, model_validators, 'inner')
577 model_schema = core_schema.model_schema(
578 cls,
579 inner_schema,
--> 580 custom_init=getattr(cls, '__pydantic_custom_init__', None),
581 root_model=False,
582 post_init=getattr(cls, '__pydantic_post_init__', None),
583 config=core_config,
584 ref=model_ref,
585 metadata=metadata,
586 )
588 schema = self._apply_model_serializers(model_schema, decorators.model_serializers.values())
589 schema = apply_model_validators(schema, model_validators, 'outer')
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:916, in _generate_md_field_schema(self, name, field_info, decorators)
906 common_field = self._common_field_schema(name, field_info, decorators)
907 return core_schema.model_field(
908 common_field['schema'],
909 serialization_exclude=common_field['serialization_exclude'],
(...)
913 metadata=common_field['metadata'],
914 )
--> 916 def _generate_dc_field_schema(
917 self,
918 name: str,
919 field_info: FieldInfo,
920 decorators: DecoratorInfos,
921 ) -> core_schema.DataclassField:
922 """Prepare a DataclassField to represent the parameter/field, of a dataclass."""
923 common_field = self._common_field_schema(name, field_info, decorators)
``````output
File ~/job/zep-proprietary/venv/lib/python3.11/site-packages/pydantic/_internal/_generate_schema.py:1114, in _common_field_schema(self, name, field_info, decorators)
1108 json_schema_extra = field_info.json_schema_extra
1110 metadata = build_metadata_dict(
1111 js_annotation_functions=[get_json_schema_update_func(json_schema_updates, json_schema_extra)]
1112 )
-> 1114 alias_generator = self._config_wrapper.alias_generator
1115 if alias_generator is not None:
1116 self._apply_alias_generator_to_field_info(alias_generator, field_info, name)
``````output
AttributeError: 'FieldInfo' object has no attribute 'deprecated'
# Provide your OpenAI key
import getpass

openai_key = getpass.getpass()
# Provide your Zep API key. See https://help.getzep.com/projects#api-keys

zep_api_key = getpass.getpass()

Initialize the Zep Chat Message History Class and initialize the Agent​

search = WikipediaAPIWrapper()
tools = [
Tool(
name="Search",
func=search.run,
description=(
"useful for when you need to search online for answers. You should ask"
" targeted questions"
),
),
]

# Set up Zep Chat History
memory = ZepCloudMemory(
session_id=session_id,
api_key=zep_api_key,
return_messages=True,
memory_key="chat_history",
)

# Initialize the agent
llm = OpenAI(temperature=0, openai_api_key=openai_key)
agent_chain = initialize_agent(
tools,
llm,
agent=AgentType.CONVERSATIONAL_REACT_DESCRIPTION,
verbose=True,
memory=memory,
)

Add some history data​

# Preload some messages into the memory. The default message window is 12 messages. We want to push beyond this to demonstrate auto-summarization.
test_history = [
{"role": "human", "content": "Who was Octavia Butler?"},
{
"role": "ai",
"content": (
"Octavia Estelle Butler (June 22, 1947 – February 24, 2006) was an American"
" science fiction author."
),
},
{"role": "human", "content": "Which books of hers were made into movies?"},
{
"role": "ai",
"content": (
"The most well-known adaptation of Octavia Butler's work is the FX series"
" Kindred, based on her novel of the same name."
),
},
{"role": "human", "content": "Who were her contemporaries?"},
{
"role": "ai",
"content": (
"Octavia Butler's contemporaries included Ursula K. Le Guin, Samuel R."
" Delany, and Joanna Russ."
),
},
{"role": "human", "content": "What awards did she win?"},
{
"role": "ai",
"content": (
"Octavia Butler won the Hugo Award, the Nebula Award, and the MacArthur"
" Fellowship."
),
},
{
"role": "human",
"content": "Which other women sci-fi writers might I want to read?",
},
{
"role": "ai",
"content": "You might want to read Ursula K. Le Guin or Joanna Russ.",
},
{
"role": "human",
"content": (
"Write a short synopsis of Butler's book, Parable of the Sower. What is it"
" about?"
),
},
{
"role": "ai",
"content": (
"Parable of the Sower is a science fiction novel by Octavia Butler,"
" published in 1993. It follows the story of Lauren Olamina, a young woman"
" living in a dystopian future where society has collapsed due to"
" environmental disasters, poverty, and violence."
),
"metadata": {"foo": "bar"},
},
]

for msg in test_history:
memory.chat_memory.add_message(
(
HumanMessage(content=msg["content"])
if msg["role"] == "human"
else AIMessage(content=msg["content"])
),
metadata=msg.get("metadata", {}),
)

Run the agent​

Doing so will automatically add the input and response to the Zep memory.

agent_chain.invoke(
input="What is the book's relevance to the challenges facing contemporary society?",
)


> Entering new AgentExecutor chain...

AI: Parable of the Sower is highly relevant to contemporary society as it explores themes of environmental degradation, social and economic inequality, and the struggle for survival in a chaotic world. It also delves into issues of race, gender, and religion, making it a thought-provoking and timely read.

> Finished chain.
{'input': "What is the book's relevance to the challenges facing contemporary society?",
'chat_history': [HumanMessage(content="Octavia Butler's contemporaries included Ursula K. Le Guin, Samuel R. Delany, and Joanna Russ.\nOctavia Butler won the Hugo Award, the Nebula Award, and the MacArthur Fellowship.\nUrsula K. Le Guin is known for novels like The Left Hand of Darkness and The Dispossessed.\nJoanna Russ is the author of the influential feminist science fiction novel The Female Man.\nMargaret Atwood is known for works like The Handmaid's Tale and the MaddAddam trilogy.\nConnie Willis is an award-winning author of science fiction and fantasy, known for novels like Doomsday Book.\nOctavia Butler is a pioneering black female science fiction author, known for Kindred and the Parable series.\nOctavia Estelle Butler was an acclaimed American science fiction author. While none of her books were directly adapted into movies, her novel Kindred was adapted into a TV series on FX. Butler was part of a generation of prominent science fiction writers in the 20th century, including contemporaries such as Ursula K. Le Guin, Samuel R. Delany, Chip Delany, and Nalo Hopkinson.\nhuman: What awards did she win?\nai: Octavia Butler won the Hugo Award, the Nebula Award, and the MacArthur Fellowship.\nhuman: Which other women sci-fi writers might I want to read?\nai: You might want to read Ursula K. Le Guin or Joanna Russ.\nhuman: Write a short synopsis of Butler's book, Parable of the Sower. What is it about?\nai: Parable of the Sower is a science fiction novel by Octavia Butler, published in 1993. It follows the story of Lauren Olamina, a young woman living in a dystopian future where society has collapsed due to environmental disasters, poverty, and violence.")],
'output': 'Parable of the Sower is highly relevant to contemporary society as it explores themes of environmental degradation, social and economic inequality, and the struggle for survival in a chaotic world. It also delves into issues of race, gender, and religion, making it a thought-provoking and timely read.'}

Inspect the Zep memory​

Note the summary, and that the history has been enriched with token counts, UUIDs, and timestamps.

Summaries are biased towards the most recent messages.

def print_messages(messages):
for m in messages:
print(m.type, ":\n", m.dict())


print(memory.chat_memory.zep_summary)
print("\n")
print("Conversation Facts: ")
facts = memory.chat_memory.zep_facts
for fact in facts:
print(fact + "\n")
print_messages(memory.chat_memory.messages)
Octavia Estelle Butler was an acclaimed American science fiction author. While none of her books were directly adapted into movies, her novel Kindred was adapted into a TV series on FX. Butler was part of a generation of prominent science fiction writers in the 20th century, including contemporaries such as Ursula K. Le Guin, Samuel R. Delany, Chip Delany, and Nalo Hopkinson.


Conversation Facts:
Octavia Butler's contemporaries included Ursula K. Le Guin, Samuel R. Delany, and Joanna Russ.

Octavia Butler won the Hugo Award, the Nebula Award, and the MacArthur Fellowship.

Ursula K. Le Guin is known for novels like The Left Hand of Darkness and The Dispossessed.

Joanna Russ is the author of the influential feminist science fiction novel The Female Man.

Margaret Atwood is known for works like The Handmaid's Tale and the MaddAddam trilogy.

Connie Willis is an award-winning author of science fiction and fantasy, known for novels like Doomsday Book.

Octavia Butler is a pioneering black female science fiction author, known for Kindred and the Parable series.

Parable of the Sower is a science fiction novel by Octavia Butler, published in 1993.

The novel follows the story of Lauren Olamina, a young woman living in a dystopian future where society has collapsed due to environmental disasters, poverty, and violence.

Parable of the Sower explores themes of environmental degradation, social and economic inequality, and the struggle for survival in a chaotic world.

The novel also delves into issues of race, gender, and religion, making it a thought-provoking and timely read.

human :
{'content': "Octavia Butler's contemporaries included Ursula K. Le Guin, Samuel R. Delany, and Joanna Russ.\nOctavia Butler won the Hugo Award, the Nebula Award, and the MacArthur Fellowship.\nUrsula K. Le Guin is known for novels like The Left Hand of Darkness and The Dispossessed.\nJoanna Russ is the author of the influential feminist science fiction novel The Female Man.\nMargaret Atwood is known for works like The Handmaid's Tale and the MaddAddam trilogy.\nConnie Willis is an award-winning author of science fiction and fantasy, known for novels like Doomsday Book.\nOctavia Butler is a pioneering black female science fiction author, known for Kindred and the Parable series.\nParable of the Sower is a science fiction novel by Octavia Butler, published in 1993.\nThe novel follows the story of Lauren Olamina, a young woman living in a dystopian future where society has collapsed due to environmental disasters, poverty, and violence.\nParable of the Sower explores themes of environmental degradation, social and economic inequality, and the struggle for survival in a chaotic world.\nThe novel also delves into issues of race, gender, and religion, making it a thought-provoking and timely read.\nOctavia Estelle Butler was an acclaimed American science fiction author. While none of her books were directly adapted into movies, her novel Kindred was adapted into a TV series on FX. Butler was part of a generation of prominent science fiction writers in the 20th century, including contemporaries such as Ursula K. Le Guin, Samuel R. Delany, Chip Delany, and Nalo Hopkinson.\nhuman: Which other women sci-fi writers might I want to read?\nai: You might want to read Ursula K. Le Guin or Joanna Russ.\nhuman: Write a short synopsis of Butler's book, Parable of the Sower. What is it about?\nai: Parable of the Sower is a science fiction novel by Octavia Butler, published in 1993. It follows the story of Lauren Olamina, a young woman living in a dystopian future where society has collapsed due to environmental disasters, poverty, and violence.\nhuman: What is the book's relevance to the challenges facing contemporary society?\nai: Parable of the Sower is highly relevant to contemporary society as it explores themes of environmental degradation, social and economic inequality, and the struggle for survival in a chaotic world. It also delves into issues of race, gender, and religion, making it a thought-provoking and timely read.", 'additional_kwargs': {}, 'response_metadata': {}, 'type': 'human', 'name': None, 'id': None, 'example': False}

Vector search over the Zep memory​

Zep provides native vector search over historical conversation memory via the ZepRetriever.

You can use the ZepRetriever with chains that support passing in a Langchain Retriever object.

retriever = ZepCloudRetriever(
session_id=session_id,
api_key=zep_api_key,
)

search_results = memory.chat_memory.search("who are some famous women sci-fi authors?")
for r in search_results:
if r.score > 0.8: # Only print results with similarity of 0.8 or higher
print(r.message, r.score)
content='Which other women sci-fi writers might I want to read?' created_at='2024-05-10T14:34:16.714292Z' metadata=None role='human' role_type=None token_count=12 updated_at='0001-01-01T00:00:00Z' uuid_='64ca1fae-8db1-4b4f-8a45-9b0e57e88af5' 0.8960460126399994

Was this page helpful?


You can also leave detailed feedback on GitHub.