Cortex Memory

Give your AI agents memory and learning capabilities. Remember context, learn from patterns, and improve over time.

Four Types of Memory

Just like human cognition, your agents get multiple memory systems

Working Memory

Short-term context for conversations. Lightning-fast access to recent interactions.

Episodic Memory

Remember past interactions. Build context over time for better personalization.

Semantic Memory

Knowledge base with vector search. Store and retrieve information semantically.

Procedural Memory

Learn patterns and optimize behavior. Improve performance over time automatically.

Enable Memory in One Line

Cortex integrates seamlessly with your existing code

from teleon import TeleonClient

client = TeleonClient(api_key="your-api-key")

@client.agent(cortex={
    # Enable multiple memory types
    "memory_types": ["working", "episodic", "semantic"],
    
    # Storage backend
    "storage": "managed",  # or "redis", "postgres"
    
    # Automatic learning
    "learning": {
        "enabled": True,
        "pattern_recognition": True
    },
    
    # Embedding model for semantic search
    "embedding_model": "text-embedding-3-small"
})
async def smart_agent(message: str, user_id: str) -> str:
    # Cortex automatically manages memory
    # - Remembers conversation history
    # - Recalls relevant past interactions
    # - Learns from user preferences
    
    response = await generate_response(message, user_id)
    return response

# That's it! Memory is automatic.
80%

Reduction in API costs with smart caching

<10ms

Memory access latency

100%

Automatic persistence

Advanced Features

Built for enterprise-scale applications

Automatic Persistence

All memory is automatically saved and restored across restarts

Vector Embeddings

Semantic search powered by state-of-the-art embedding models

Pattern Recognition

Automatically identify and learn from recurring patterns

Cost Optimization

Smart caching reduces LLM API calls by up to 80%

Privacy Controls

Fine-grained control over what data is stored and for how long

Multi-Backend

Redis, PostgreSQL, or bring your own storage solution

Real-World Applications

See how companies use Cortex

Customer Support

Spotify uses Cortex to remember customer interactions across channels. Support agents have full context of past issues, reducing resolution time by 60%.

Impact:60% faster resolution

E-commerce Personalization

Shopify merchants use Cortex to remember customer preferences and shopping patterns. Semantic memory powers product recommendations.

Impact:2.5x increase in conversion

Data Analysis

Financial firms use Cortex to build knowledge bases from documents. Semantic memory enables natural language queries across terabytes of data.

Impact:90% faster research

Healthcare Assistants

Medical platforms use Cortex to remember patient history and medication records. Procedural memory learns from clinical patterns.

Impact:50% reduction in errors

How Cortex Works

Under the hood: A sophisticated memory architecture

1

Capture

Every interaction is automatically captured and stored with metadata

2

Embed

Semantic content is converted to vector embeddings for similarity search

3

Retrieve

Relevant memories are retrieved based on context and similarity

4

Learn

Patterns are identified and used to optimize future interactions

Give your agents memory today

Start building smarter AI agents with Cortex