> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cognee.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Migrate from Mem0
> Move from Mem0 to Cognee
This guide shows how to export your existing Mem0 memories and import them into Cognee.
Need help with a large migration? [Chat with us](https://calendly.com/vasilije-topoteretes/)
| Capability | **Cognee** | **Mem0** |
| ---------------------- | ------------------------------------------------------------------------------- | --------------------------------------------------------------- |
| **Primary Storage** | Built-in knowledge graph + vector store, with scalable options | Vectors (optional graphs) |
| **Data Ingestion** | Creates memories from any data (chats, documents, etc.) allows batch processing | Creates memories from conversations |
| **Entity Extraction** | Multi-layer semantic connections that tie the graph together more densely | Lean with basic entity nets |
| **Memory Enrichment** | `memify()` derives new facts and enriches existing graphs over time | Not available |
| **Ontology Support** | RDF/OWL ontologies and Pydantic models for custom schemas | Not available |
| **Temporal Awareness** | Time-aware entity extraction and temporal search type | Not available |
| **Feedback Loop** | User feedback during search is stored to improve reasoning | Not available |
| **Search Types** | 15 retrieval types available to optimize for use case | Vector similarity with filters (basic graph queries if enabled) |
| **Data Isolation** | Dataset-scoped permissions and per-dataset storage | Isolation through user\_id, agent\_id, and run\_id |
| **API Style** | Asynchronous (async/await) | Synchronous (sync) |
## Data Migration Script
Export from Mem0 and import to Cognee in one script:
```python theme={null}
import asyncio
from mem0 import Memory
import cognee
from cognee.api.v1.search import SearchType
async def migrate():
# Export from Mem0
m = Memory()
mem0_data = m.get_all(user_id="alice")
# Reset Cognee (optional, for clean start)
await cognee.prune.prune_data()
await cognee.prune.prune_system(metadata=True)
# Import to Cognee
for memory in mem0_data.get("results", []):
content = memory.get("memory", "")
if content:
await cognee.add(content, dataset_name="alice")
# Build knowledge graph
await cognee.cognify()
# Verify
result = await cognee.search(
query_type=SearchType.GRAPH_COMPLETION,
query_text="What do you know?"
)
print(result)
asyncio.run(migrate())
```
## API Mapping
Use this table to translate Mem0 operations to Cognee equivalents:
| **Operation** | **Cognee** | **Mem0** |
| ---------------- | ------------------------------------------------------------------------------------ | -------------------------------------- |
| Add text | `await cognee.add("text", dataset_name="alice")` | `m.add("text", user_id="alice")` |
| Add file | `await cognee.add("/path/to/file.pdf", dataset_name="docs")` | *Not supported* |
| Add URL | `await cognee.add("https://example.com", dataset_name="web")` | *Not supported* |
| Build graph | `await cognee.cognify()` | Automatic (if graph\_store configured) |
| Search | `await cognee.search(query_text="query", query_type=SearchType.GRAPH_COMPLETION)` | `m.search("query", user_id="alice")` |
| Get all | `await cognee.datasets.list_data(dataset_id)` | `m.get_all(user_id="alice")` |
| Delete user data | `await cognee.datasets.empty_dataset(dataset_id)` | `m.delete_all(user_id="alice")` |
| Reset all | `await cognee.prune.prune_data()` + `await cognee.prune.prune_system(metadata=True)` | `m.reset()` |
To get the `dataset_id`, use `await cognee.datasets.list_datasets()`.
### Code Migration Examples
Below are side-by-side comparisons of the most common memory operations in Cognee and Mem0.
#### Adding Data
```python Cognee theme={null}
import cognee
# Single text
await cognee.add("User loves pizza", dataset_name="alice")
# Conversation messages
messages = [
{"role": "user", "content": "I love pizza"},
{"role": "assistant", "content": "Noted!"}
]
await cognee.add(messages, dataset_name="alice")
# Build graph after adding
await cognee.cognify()
```
```python Mem0 theme={null}
from mem0 import Memory
m = Memory()
# Single text
m.add("User loves pizza", user_id="alice")
# Conversation messages
messages = [
{"role": "user", "content": "I love pizza"},
{"role": "assistant", "content": "Noted!"}
]
m.add(messages, user_id="alice")
```
#### Searching
```python Cognee theme={null}
import cognee
from cognee.api.v1.search import SearchType
# Basic search
answer = await cognee.search(
query_text="What does user like?",
query_type=SearchType.GRAPH_COMPLETION,
)
# With dataset filter (like user_id)
answer = await cognee.search(
query_text="What does user like?",
query_type=SearchType.GRAPH_COMPLETION,
datasets=["alice"],
)
```
```python Mem0 theme={null}
from mem0 import Memory
m = Memory()
# Basic search
results = m.search("What does user like?", user_id="alice")
for r in results["results"]:
print(r["memory"])
```
#### Deleting
```python Cognee theme={null}
import cognee
# Delete an entire dataset
datasets = await cognee.datasets.list_datasets()
dataset_id = datasets[0].id
await cognee.datasets.empty_dataset(dataset_id)
# Full reset
await cognee.prune.prune_data()
await cognee.prune.prune_system(metadata=True)
```
```python Mem0 theme={null}
from mem0 import Memory
m = Memory()
# Delete all for user
m.delete_all(user_id="alice")
# Reset everything
m.reset()
```
For a detailed comparison of AI memory architectures and benchmark results, see [Deep Dive: AI Memory Comparison](https://www.cognee.ai/blog/deep-dives/competition-comparison-form-vs-function).
## Next Steps
Set up Cognee in your environment
Configure LLM, embedding, and storage backends
Learn about supported data sources
Explore available search types