Skip to main content
Store user preferences, conversation history, and facts in your AI application’s memory system for intelligent recall and personalization. Store information and conversations in your memory system.

​
Basic Usage

  • Python SDK
  • JavaScript SDK
from gravixlayer import GravixLayer

client = GravixLayer()
memory = client.memory

# Add simple text
result = memory.add("I love pizza", user_id="alice")
print(f"Added memory: {result['results'][0]['memory']}")
print(f"Memory ID: {result['results'][0]['id']}")

# Add with metadata
result = memory.add("User prefers dark mode", user_id="alice", metadata={"type": "preference"})
print(f"Added preference: {result['results'][0]['memory']}")
print(f"Metadata: {result['results'][0]['metadata']}")

# Get all memories to verify
all_memories = memory.get_all(user_id="alice")
print(f"\nTotal memories for alice: {len(all_memories['results'])}")
for i, mem in enumerate(all_memories['results'], 1):
    print(f"{i}. {mem['memory']}")
    if mem.get('metadata'):
        print(f"   Metadata: {mem['metadata']}")
Expected Output:
Added memory: I love pizza
Memory ID: b355d0d2-3eaa-4bc6-a61b-48ee615279bf
Added preference: User prefers dark mode
Metadata: {'type': 'preference'}

Total memories for alice: 2
1. I love pizza
2. User prefers dark mode
   Metadata: {'type': 'preference'}

​
Add Conversations

Store entire conversations and let AI extract key memories:
  • Python SDK
  • JavaScript SDK
from gravixlayer import GravixLayer

client = GravixLayer()
memory = client.memory

# Store a conversation with AI inference
conversation = [
    {"role": "user", "content": "I'm planning to watch a movie tonight. Any recommendations?"},
    {"role": "assistant", "content": "How about thriller movies? They can be quite engaging."},
    {"role": "user", "content": "I'm not a big fan of thriller movies but I love sci-fi movies."},
    {"role": "assistant", "content": "Got it! I'll avoid thriller recommendations and suggest sci-fi movies in the future."}
]

result = memory.add(conversation, user_id="alice", infer=True, metadata={"type": "conversation"})
print(f"AI extracted {len(result['results'])} memories from conversation:")

for i, extracted_memory in enumerate(result['results'], 1):
    print(f"{i}. {extracted_memory['memory']}")
    print(f"   ID: {extracted_memory['id']}")
    if extracted_memory.get('metadata'):
        print(f"   Metadata: {extracted_memory['metadata']}")

# Verify by searching for movie preferences
search_results = memory.search("movie preferences", user_id="alice")
print(f"\nFound {len(search_results['results'])} movie-related memories:")
for result in search_results['results']:
    print(f"- {result['memory']}")
Expected Output:
AI extracted 2 memories from conversation:
1. User prefers sci-fi movies
   ID: c455d0d2-3eaa-4bc6-a61b-48ee615279bf
   Metadata: {'type': 'conversation'}
2. User dislikes thriller movies
   ID: d755d0d2-3eaa-4bc6-a61b-48ee615279bf
   Metadata: {'type': 'conversation'}

Found 2 movie-related memories:
- User prefers sci-fi movies
- User dislikes thriller movies

​
Custom Configuration (Optional)

The memory system uses smart defaults, but you can customize it for specific needs:

​
Configuration Parameters

What each setting does:
  • embedding_model - How text gets converted to searchable vectors
  • inference_model - AI model that extracts memories from conversations
  • index_name - Where memories are stored (like folders)
  • cloud_provider - Where your data is hosted
  • Python SDK
  • JavaScript SDK
from gravixlayer import GravixLayer

client = GravixLayer()
memory = client.memory

# Configure for multilingual app with organized storage
memory.switch_configuration(
    embedding_model="microsoft/multilingual-e5-large",  # Supports 100+ languages
    inference_model="qwen/qwen-2.5-vl-7b-instruct",    # Better context understanding
    index_name="user_preferences",                      # Organized storage
    cloud_provider="GCP",                               # Google Cloud hosting
    region="us-east1"                                   # Specific region
)

# Now works with any language
result1 = memory.add("El usuario prefiere pizza", user_id="alice")
result2 = memory.add("L'utilisateur aime le café", user_id="alice")
result3 = memory.add("用户喜欢寿司", user_id="alice")

print("Added multilingual memories:")
print(f"Spanish: {result1['results'][0]['memory']}")
print(f"French: {result2['results'][0]['memory']}")
print(f"Chinese: {result3['results'][0]['memory']}")

# Check current configuration
config = memory.get_current_configuration()
print(f"\nCurrent configuration:")
print(f"Embedding model: {config['embedding_model']}")
print(f"Inference model: {config['inference_model']}")
print(f"Index name: {config['index_name']}")
print(f"Cloud provider: {config['cloud_provider']}")
print(f"Region: {config['region']}")

# Search works across all languages
search_results = memory.search("food preferences", user_id="alice")
print(f"\nFound {len(search_results['results'])} food-related memories:")
for result in search_results['results']:
    print(f"- {result['memory']}")
Expected Output:
🔄 Switched embedding model to: microsoft/multilingual-e5-large
🔄 Switched inference model to: qwen/qwen-2.5-vl-7b-instruct
🔄 Switched to database: user_preferences
🔄 Switched cloud provider to: GCP
🔄 Switched region to: us-east1
✅ Configuration updated successfully

Added multilingual memories:
Spanish: El usuario prefiere pizza
French: L'utilisateur aime le café
Chinese: 用户喜欢寿司

Current configuration:
Embedding model: microsoft/multilingual-e5-large
Inference model: qwen/qwen-2.5-vl-7b-instruct
Index name: user_preferences
Cloud provider: GCP
Region: us-east1

Found 3 food-related memories:
- El usuario prefiere pizza
- L'utilisateur aime le café
- 用户喜欢寿司
⌘I