extremis

Memory that gets smarter the more your agent uses it.

extremis is an open-source memory layer for AI agents. It handles embedding, storage, retrieval ranking, and consolidation — so you don't have to.

---

The fastest path

Change one import. Get persistent, learning memory for free.

Claude (Anthropic)OpenAIManual API

from extremis.wrap import Anthropic  # (1)
from extremis import Extremis

client = Anthropic(api_key="sk-ant-...", memory=Extremis())

# Your existing code — unchanged
response = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[{"role": "user", "content": "What's my name?"}]
)
# Memory recalled before call, saved after. Nothing else to do.

1. Drop-in for anthropic.Anthropic. Every other method passes through unchanged.

from extremis.wrap import OpenAI
from extremis import Extremis

client = OpenAI(api_key="sk-...", memory=Extremis())
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "What did we discuss last time?"}]
)

from extremis import Extremis

mem = Extremis()

# Store
mem.remember("User is building a WhatsApp AI", conversation_id="c1")

# Recall — every result explains why it ranked
results = mem.recall("what is the user building?")
for r in results:
    print(r.memory.content)
    print(r.reason)  # "similarity 0.91 · score +2.0 · used 5× · 3d old"

# Feedback — useful memories surface first over time
mem.report_outcome([results[0].memory.id], success=True)

---

What makes it different

	extremis	Mem0	LangChain	Raw vectors
RL-scored retrieval	✅	❌	❌	❌
Memory explains itself	✅	❌	❌	❌
Knowledge graph	✅	❌	❌	❌
Drop-in LLM wrapper	✅	❌	❌	❌
Backend-agnostic	✅ 4 backends	Cloud only	Manual	—
MCP server (Claude)	✅	❌	❌	❌
Self-hostable	✅	❌	✅	✅
Open source (MIT)	✅	Partial	✅	✅

---

Install

pip3.11 install extremis                    # core
pip3.11 install "extremis[wrap-anthropic]"  # + Claude wrapper
pip3.11 install "extremis[wrap-openai]"     # + OpenAI wrapper
pip3.11 install "extremis[mcp]"             # + Claude Desktop MCP

Python 3.11+ required

If pip install says "no matching distribution found", your pip points to Python 3.9. Run python3 --version to check. Fix: brew install python@3.11 then use pip3.11.

---

Next steps

• :material-clock-fast: Quickstart

  Get memory working in 5 minutes

• :material-swap-horizontal: Wrap your existing app

  One import change for Claude or OpenAI

• :material-brain: Core concepts

  How layers, RL scoring, and consolidation work

• :material-server: Deploy to Render

  Hosted server with persistent Postgres memory