Graph Memory

Fresh

Graph Memory establishes connections between data entities, creating a network of interconnected information for more contextually relevant search outcomes.

How It Works

1. Mem0 constructs a graph representation of entities automatically
2. Vector search identifies the top semantic matches (with optional reranking)
3. Graph relations accompany results for additional context

graph LR
    A[Joseph] -->|name| B[Joseph]
    A -->|city| C[Seattle]
    A -->|job| D[Software Engineer]

Adding Memories with Graph Memory

Enable Graph Memory using the enable_graph=True parameter.

Python

from mem0 import MemoryClient

client = MemoryClient(
    api_key="your-api-key",
    org_id="your-org-id",
    project_id="your-project-id"
)

messages = [
    {"role": "user", "content": "My name is Joseph"},
    {"role": "assistant", "content": "Hello Joseph, nice to meet you!"},
    {"role": "user", "content": "I'm from Seattle and I work as a software engineer"}
]

client.add(messages, user_id="joseph", enable_graph=True)

JavaScript

import { MemoryClient } from "mem0";

const client = new MemoryClient({
  apiKey: "your-api-key",
  org_id: "your-org-id",
  project_id: "your-project-id"
});

const messages = [
  { role: "user", content: "My name is Joseph" },
  { role: "assistant", content: "Hello Joseph, nice to meet you!" },
  { role: "user", content: "I'm from Seattle and I work as a software engineer" }
];

await client.add({
  messages,
  user_id: "joseph",
  enable_graph: true
});

INFO

Graph memory add responses are not immediately available since the operation processes asynchronously. Use get_all() to retrieve memories with graph metadata.

Searching with Graph Memory

Python

results = client.search(
    "what is my name?",
    user_id="joseph",
    enable_graph=True
)

JavaScript

const results = await client.search({
  query: "what is my name?",
  user_id: "joseph",
  enable_graph: true
});

Search Response

{
  "results": [
    {
      "id": "4a5a417a-...",
      "memory": "Name is Joseph",
      "user_id": "joseph",
      "categories": ["personal_details"],
      "score": 0.3621
    }
  ],
  "relations": [
    {
      "source": "joseph",
      "source_type": "person",
      "relationship": "name",
      "target": "joseph",
      "target_type": "person",
      "score": 0.39
    }
  ]
}

TIP

Results reflect vector search order (optionally reranked). Graph Memory augments responses by adding related entities in the relations array without reordering vector results.

Retrieving All Memories with Graph

Python

memories = client.get_all(
    filters={"AND": [{"user_id": "joseph"}]},
    enable_graph=True
)

JavaScript

const memories = await client.getAll({
  filters: { AND: [{ user_id: "joseph" }] },
  enable_graph: true
});

WARNING

get_all() requires filters to be specified.

Project-Level Graph Memory

Set graph memory at the project level so all operations use it automatically:

Python

client.project.update(enable_graph=True)

# Now all add() calls use graph memory by default
client.add(messages, user_id="joseph")

JavaScript

await client.project.update({ enable_graph: true });

// Now all add() calls use graph memory by default
await client.add({ messages, user_id: "joseph" });

Best Practices

• Use Graph Memory when understanding entity relationships matters
• Performance improves with substantial interaction history
• Graph Memory suits long-term assistants managing evolving information
• Expect slightly longer response times for large memory repositories