Open Source · MIT Licensed

Memory for
AI Agents

Your agents forget everything between sessions. NeuroCache gives them persistent, hierarchical memory—reducing token costs by 70-90%.

70-90%cost reduction
3lines to integrate
agent.py
# Add memory in 3 lines
from neurocache import NeuroCache
cache = NeuroCache(agent="my-agent")
await cache.remember(context)
# Your agent now remembers
✓Long-term memory
✓Working memory
✓Cross-session recall
live demo
ready
> cache.recall(""|)

Built for production

Not another toy. NeuroCache is designed for real AI agents processing millions of tokens.

CORE

Hierarchical Memory Architecture

Five memory types modeled after human cognition: sensory, short-term, working, long-term, and episodic. Each with configurable decay rates and consolidation rules.

Sensory
Short-term
Working
Long-term
Episodic

70-90% Cost Reduction

Stop sending the same context repeatedly. NeuroCache intelligently caches and retrieves only what's needed.

$2.4k
avg. monthly savings per agent

3-Line Integration

Native support for LangChain, LlamaIndex, and any LLM. Python SDK and REST API.

Production Ready

PostgreSQL + pgvector, Redis caching, Kubernetes-ready. 99.9% uptime SLA.

How it works

Four steps to agents that actually remember.

01
01

Connect

Import the SDK and initialize with your agent ID. One line.

cache = NeuroCache(agent='id')
02
02

Store

Call remember() with any context. We handle the rest.

await cache.remember(conversation)
03
03

Recall

Retrieve relevant memories automatically or on-demand.

context = await cache.recall(query)
04
04

Save

Watch your token costs drop as memory kicks in.

# 70-90% fewer tokens sent
Average integration time: 15 minutes

Simple pricing

Start free. Scale when you need to.

Open Source

Free

Self-hosted, MIT licensed

Perfect for experimentation and small projects. Full access to core features with community support.

  • Unlimited local usage
  • All memory types
  • Community Discord
  • Full source access
View on GitHub

Startup

$29/mo

For growing teams

Ideal for startups building their first AI-powered products. Generous limits and team collaboration.

  • 2M tokens/month
  • Up to 5 team members
  • Email support
  • Usage analytics
Early Access

Cloud

$99/mo

Managed service

Production-ready infrastructure for scaling AI applications. Priority support and advanced features.

  • 10M tokens/month
  • Unlimited team members
  • Priority support
  • Advanced analytics

Enterprise

Custom

On-prem or dedicated cloud

For organizations with specific security, compliance, or scale requirements. White-glove onboarding.

  • Unlimited tokens
  • 24/7 dedicated support
  • SOC2, HIPAA, GDPR
  • Custom SLAs

Ready to give your agents memory?

Join the developers building AI that actually learns.

Star on GitHub·View source·Docs