ZwartifyOS

🧠

Agent Memory System

Bring your own memory provider. Redis for short-term, Pinecone for long-term. Agents that remember and learn.

Currently in development. Configure your own Redis or Pinecone instance to enable persistent, context-aware agent conversations.

View Documentation →Try Playground

🚧

Memory System Coming Soon

The memory system is currently in development. You'll be able to connect your own Redis or Pinecone instance for agent memory. Stay tuned for updates!

Memory Types

⚡

Short-Term Memory

Fast, ephemeral storage for conversation context

Redis

Use Cases:

→Session-based conversations
→Temporary context storage
→Real-time message history
→Conversation state management

Benefits:

Ultra-fast read/writeLow latencyPerfect for active sessionsAutomatic expiration

🧠

Long-Term Memory

Persistent, searchable memory for agent learning

Pinecone

Use Cases:

→Cross-session learning
→Agent knowledge base
→Semantic search
→Pattern recognition

Benefits:

Vector search capabilitiesSemantic understandingLong-term retentionScalable storage

Memory Features

🔌

Bring Your Own Memory

Use Redis, Pinecone, or any compatible memory provider. No vendor lock-in.

⚡

Dual Memory System

Short-term for conversations, long-term for learning. Best of both worlds.

💭

Context-Aware Agents

Agents remember previous conversations and learn from interactions.

🔍

Semantic Search

Find relevant memories using meaning, not just keywords.

🤖

Automatic Management

Memory is automatically stored and retrieved based on relevance.

🔒

Privacy Control

Control what agents remember and how long data is retained.

How It Works

1️⃣

Choose Your Memory Provider

Select Redis for short-term or Pinecone for long-term memory (or both!)

# Short-term memory (Redis) REDIS_URL=redis://localhost:6379 REDIS_PASSWORD=your_password # Long-term memory (Pinecone) PINECONE_API_KEY=your_api_key PINECONE_ENVIRONMENT=us-east-1 PINECONE_INDEX=zwartify-memory

2️⃣

Configure in Environment

Set environment variables for your chosen memory provider

# .env.local MEMORY_TYPE=dual # or 'redis', 'pinecone', 'none' ENABLE_MEMORY=true

3️⃣

Agents Use Memory Automatically

Once configured, agents automatically store and retrieve memories

// No code changes needed! // Agents automatically use memory when enabled

With vs Without Memory

Short-Term Memory

Without Memory:

Context lost between sessions

With Memory:

Conversations persist during active sessions

Long-Term Memory

Without Memory:

Agents start fresh every time

With Memory:

Agents learn and remember across sessions

User Experience

Without Memory:

Repeat information every interaction

With Memory:

Agents remember preferences and context

Agent Intelligence

Without Memory:

Static responses

With Memory:

Evolving, learning behaviour

Technical Details

Redis (Short-Term)

→Stores conversation context in memory
→Fast read/write operations
→Automatic expiration (configurable TTL)
→Perfect for session-based storage
→Low latency, high throughput

Pinecone (Long-Term)

→Vector database for semantic search
→Stores embeddings of conversations
→Finds relevant memories by meaning
→Persistent across sessions
→Scales to millions of memories

Ready to Enable Memory?

Once the memory system is available, configure your Redis or Pinecone instance and agents will automatically use memory for context-aware conversations.

View Documentation Try Playground