Persistent memory that remembers everything
Memphora enables AI agents to recall past interactions, store user preferences, learn from experiences, and maintain complex relationships between memories.
Memphora AI Assistant enabled chatbot
With persistent memory
Starting conversation...
Powerful Dashboard
Complete dashboard interface with real-time insights and analytics
Overview
Real-time analytics and insights
Memory Growth
API Usage
Response Time Trend
Search Accuracy Trend
API Keys
API Key
Created 11/7/2025, 9:06:33 PM
Recent Activity
Created memory
john@example.com
2 mins ago
Searched memories
sarah@company.com
5 mins ago
Updated memory
mike@test.com
12 mins ago
Features That Set Us Apart
Advanced capabilities and exclusive features that make Memphora the most powerful memory layer for AI agents.
3-Layer Caching System
Exclusive multi-level caching architecture (L1 hot cache, L2 warm cache, L3 persistent cache) delivers ultra-fast retrieval with optimized performance. No other solution offers this advanced caching strategy.
Advanced Context Compression
Intelligent compression engine that achieves significant token reduction while preserving meaning and relevance. Our proprietary compression algorithms go beyond basic summarization.
AI-Powered Memory Merging
Automatically detect and merge similar memories, resolve contradictions, and maintain consistency using advanced AI models. Fully automated conflict resolution that keeps your memory clean.
Graph-Based Memory Relationships
Link memories with semantic relationships (related, contradicts, supports, extends) to build a knowledge graph. Navigate memory clusters and discover connections that other solutions miss.
Analytics & Observability
Comprehensive analytics, memory growth tracking, API usage statistics, and performance metrics. Built-in observability with metrics and audit logs.
Production-Ready Out of the Box
Enterprise-grade security, observability, health checks, Docker deployment, and rate limiting included. No additional configuration needed for production deployment.
Multi-Factor Memory Scoring
Advanced scoring system that considers recency, frequency, relevance, and importance to rank memories intelligently. Smarter retrieval than basic similarity search.
Optimized Retrieval Engine
Better accuracy than standard vector search with filtering, scoring, and context-aware retrieval. Our optimized engine delivers more relevant results with lower latency.
Memphora Capabilities
Powerful features that make your AI agents smarter, faster, and more cost-effective.
Persistent Memory Storage
Hybrid storage using vector database and relational database for reliable, scalable memory management.
Advanced Search & Reranking
Semantic search with external reranking (Cohere/Jina) for 20-30% better relevance. Auto-fallback to built-in ensemble ranking.
Advanced Metadata Filtering
Complex AND/OR/NOT queries with 12+ operators ($gt, $in, $regex, etc.) for precise memory filtering.
Context Compression
Significant token reduction through intelligent compression while preserving meaning and relevance.
3-Layer Caching
Multi-level caching system (L1 hot cache, L2 warm cache, L3 persistent cache) for ultra-fast retrieval and optimized performance.
Memory Merging
Automatically merge similar memories, detect contradictions, and resolve conflicts using advanced AI models.
Production Ready
Enterprise-grade security, observability, health checks, and Docker deployment out of the box.
Analytics & Metrics
Comprehensive analytics, memory growth tracking, API usage statistics, and performance metrics via API endpoints.
Multi-Agent Support
Native CrewAI and AutoGen integrations with shared crew memory, per-agent namespaces, and escalation tracking.
Group Chat
Multi-user conversations with shared and private memories. Built for team collaboration.
Framework Integrations
Native support for Vercel AI SDK, LangChain, LlamaIndex, CrewAI, AutoGen, Zapier, and more.
Multi-LLM Support
Supports multiple LLM providers for text tasks, multimodal processing, and seamless integration with popular AI models.
Memphora Features
New development primitives that help simplify your code and allow you to build more features, faster.
Autosave for Application State
Memphora captures the complete state of your memories (variables, relationships, metadata) so you get the benefits of a state machine, without maintaining complex state machine code.
Documentation →from memphora_sdk import Memphora
memory = Memphora(
user_id="user123",
api_key="your-api-key-here"
)
# Store memories - state is automatically persisted
memory.store("User loves Python programming")
memory.store("User works at Google")
# Search with automatic state management
results = memory.search("What does the user love?")Advanced Search with External Reranking
Native semantic search with external reranking support (Cohere/Jina) for 20-30% better relevance. Advanced metadata filtering with AND/OR/NOT operators. Auto-fallback to built-in ranking if no API keys provided.
Documentation →# Search with external reranking (NEW!)
results = memory.search(
query="user preferences",
limit=10,
rerank=True,
rerank_provider="auto" # Cohere or Jina
)
# Advanced metadata filtering (NEW!)
results = memory.search(
query="programming",
metadata_filter={
"AND": [
{"category": "work"},
{"priority": {"$gte": 5}},
{"NOT": {"archived": True}}
]
}
)Context Compression & 3-Layer Caching
Intelligent compression reduces token usage while maintaining relevance. Combined with our 3-layer caching system (L1 hot cache, L2 warm cache, L3 persistent cache) for ultra-fast responses. An outage or restart won't prevent your compressed context from being available.
Documentation →# Automatic compression with 3-layer caching
memory = Memphora(
user_id="user123",
api_key="your-api-key-here",
auto_compress=True,
max_tokens=500
)
# Get optimized context (uses 3-layer cache)
context = memory.get_context(
query="user information",
limit=20
)Multi-Agent & Group Chat Support
Track memories by agent_id and run_id for multi-agent systems (AutoGPT, CrewAI). Group chat with shared and private memories. Perfect for team collaboration and complex agent workflows.
Documentation →# Multi-agent support
memory = Memphora(
user_id="user123",
api_key="your-api-key-here"
)
# Store memory for a specific agent
memory.store_agent_memory(
agent_id="research_agent",
content="PyTorch is popular",
run_id="run_001",
metadata={"task": "Research frameworks"}
)
# Search agent memories
results = memory.search_agent_memories(
agent_id="research_agent",
query="What frameworks are popular?",
run_id="run_001"
)
# Group chat - store shared memory
memory.store_group_memory(
group_id="ai_team_001",
content="Let's use FastAPI for the API",
metadata={"author": "alice"}
)
# Search group memories
group_results = memory.search_group_memories(
group_id="ai_team_001",
query="What did we decide about the API?",
limit=5
)One-Line Integration
Simple SDK with decorator support. Native integrations with LangChain, LlamaIndex, and AutoGPT. Drop-in replacement for any LLM framework with automatic memory management.
Documentation →from memphora_sdk import Memphora
memory = Memphora(
user_id="user123",
api_key="your-api-key-here"
)
@memory.remember
def chat(message: str, memory_context: str = "") -> str:
"""Your chatbot with automatic memory!"""
return your_ai_model(f"Context: {memory_context}\n\n{message}")Code Examples
See how easy it is to integrate Memphora into your application.
Quick Start
Get started with Memphora in seconds
from memphora_sdk import Memphora
# Initialize with your API key
memory = Memphora(
user_id="user123",
api_key="your-api-key-here"
)
# Store memories
memory.store("I love Python programming")
memory.store("I work at Google")
# Search memories
results = memory.search("What do I love?")
for mem in results:
print(mem['content'])Decorator Pattern
Automatic memory integration with decorators
from memphora_sdk import Memphora
memory = Memphora(
user_id="user123",
api_key="your-api-key-here"
)
@memory.remember
def chat(message: str, memory_context: str = "") -> str:
"""Your chatbot with automatic memory!"""
return your_ai_model(
f"Context: {memory_context}\n\n{message}"
)
# Use it - memory is automatic!
response = chat("Hello, what's my favorite language?")LangChain Integration
Seamless integration with LangChain
from memphora_sdk import MemphoraLangChain
from langchain.chains import ConversationChain
# Create LangChain-compatible memory
memory = MemphoraLangChain.create_memory(
user_id="user123",
api_key="your-api-key-here"
)
# Use with your chain
chain = ConversationChain(
llm=your_llm,
memory=memory
)Use Cases
Memphora is perfect for any application that needs persistent memory and context awareness.
AI Chatbots
Build chatbots that remember every conversation, user preference, and context across sessions. Perfect for customer support, personal assistants, and conversational AI.
- Persistent context
- User preferences
- Conversation history
- Personalized responses
Healthcare AI
Maintain patient context, treatment history, and medical preferences while ensuring HIPAA compliance and data security.
- Patient history
- Treatment tracking
- Compliance
- Secure storage
EdTech Platforms
Track learning progress, remember what students know, and personalize educational content. Adaptive learning systems that evolve with each student.
- Learning progress
- Knowledge tracking
- Adaptive content
- Progress analytics
E-Commerce
Remember customer preferences, purchase history, and browsing patterns. Provide personalized recommendations and shopping experiences.
- Purchase history
- Preferences
- Recommendations
- Personalized offers
Enterprise AI
Build enterprise AI applications that maintain context across departments, remember business rules, and provide consistent experiences.
- Business context
- Department memory
- Consistent responses
- Enterprise security
Developer Tools
Integrate memory into developer tools, IDEs, and coding assistants. Remember project context, coding patterns, and developer preferences.
- Project context
- Code patterns
- Developer preferences
- IDE integration
Join the PRO Waitlist
Get 1 year of Memphora PRO for free for the first 30 joiners by signing up with your company email
10,000 Retrieval API Requests
Per month included
50,000 Memories Storage
Per month included
Priority Support
Get help when you need it
Advanced Analytics
Deep insights dashboard
Custom Integrations
Tailored to your stack
Early Access
First to try new features
Ready to build with Memphora?
Join developers building smarter AI applications with persistent memory. Get started in minutes with our simple SDK.