xMem
xMem

Memory Orchestrator for LLMs

Instantly supercharge your LLM apps with xmem: a hybrid memory, that combines long-term knowledge and real-time context for smarter, more relevant AI.

Get Started Free

Memory System Overview

Real-time memory orchestration

Live Demo
Memory Distribution
Context Relevance
Total Memories
14,582
Avg Context
4.2 KB
Retrieval
42ms
Sessions
187

LLMs forget. Your users notice.

Stop losing context and knowledge between sessions. xmem orchestrates both persistent and session memory for every LLM call—so your AI is always relevant, accurate, and up-to-date.

LLM forgot your last conversation
Lost project or team context
Wasting time repeating yourself
Persistent memory for every user
"Remind me what we discussed?"
"Who are you again?"
"Sorry, I lost that info."
"Welcome back, Alex!"
"Here's your project summary."
"Let's pick up where you left off."

How It Works

Long-Term Memory

Store and retrieve knowledge, notes, and documents with vector search.

Session Memory

Track recent chats, instructions, and context for recency and personalization.

RAG Orchestration

Automatically assemble the best context for every LLM callno manual tuning needed.

Data Flow

User
Sends prompt
Session Store
In-memory, MongoDB
Orchestrator
Vector DB
Qdrant, ChromaDB, Pinecone
LLM Provider
Llama.cpp, Ollama, OpenAI

Why xmem?

Never Lose Knowledge

Persistent memory ensures user knowledge and context are always available.

Boost LLM Accuracy

Orchestrated context makes every LLM response more relevant and precise.

Open-Source First

Works with any open-source LLM (Llama, Mistral, etc.) and vector DB.

Effortless Integration

Easy API and dashboard for seamless integration and monitoring.

Intelligent Memory Management

Vector Database Integration

Semantic search and retrieval

Active
Embeddings
768d
Chunks
2.4k
Avg. Similarity
0.82

Memory Orchestration

Real-time context assembly

Processing
Context Size
8.2 KB
Sources
3+
Latency
38ms

Easy Integration

memory-orchestration.ts
const
orchestrator =
new
xmem({
vectorStore:
chromadb
,
sessionStore:
in-memory
,
llmProvider:
mistral
});
const
response =
await
orchestrator.query({
input:
"Tell me about our previous discussion"
});