Beta · Hosted memory for AI agents

Memory that
works like yours.

The first agent memory layer with a wake phase and a sleep phase. Your agent doesn't just retrieve facts — it consolidates, abstracts patterns, and surfaces what it figured out while you were idle.

“The agent that wakes up tomorrow is exactly the agent that went to sleep last night.” The flaw SCM fixes

Sleep is when memory consolidates.

When you sleep, your hippocampus replays the day's experiences and your cortex abstracts patterns from them. You wake up with a better version of yesterday's understanding — not the same one.

Every other agent memory product — vector stores, RAG layers, context windows — only models the wake phase. They store facts and retrieve facts. They never think between sessions.

SCM models both phases. Once per night, in the user's configured timezone, your agent runs a real consolidation cycle: schemas form, contradictions resolve, low-value memories fade. The next time the user comes back, the agent has grown.

Phase 1 — Wake

Selective attention

While the user is active, SCM behaves like prefrontal working memory: bounded, cue-driven, contradiction-aware.

  • Spreading activation across a concept graph
  • Bounded ~7-item working memory
  • Contradiction versioning — "I left Northstar"
Phase 2 — Sleep

Consolidation + dreams

At the user's bedtime, in their timezone, SCM runs NREM consolidation and REM schema extraction. Once per night.

  • NREM: rehearsal, importance reweighting, forgetting
  • REM: pattern abstraction across the day
  • Wake summary: a narrative of what was learned
484×
Faster ANN retrieval at 100K concepts vs the legacy O(n) cosine scan.
16/16
Brutal scenarios passing — multi-day recall, contradictions, idle-fired wake summaries.
Sleep cycle per night, in your user's local time. Not every five minutes.
LLM markup. Bring your own provider key — we run the memory.
How it works

A complete memory pipeline, not just a vector DB.

Six modules, one HTTP API. Every layer is exposed and configurable — but you don't have to touch any of them to get the default product behavior.

01 / ENCODING

Selective attention

An attention gate filters incoming text — "hmm okay" gets dropped; "I'm allergic to peanuts" gets encoded with high importance. Concepts go in, noise stays out.

02 / RETRIEVAL

Spreading activation

Cue-driven graph propagation, not raw vector similarity. Asks "where do I work?" returns "I'm at Atlas Labs" even though those tokens never overlap.

03 / VERSIONING

Contradiction-safe

"I left Northstar, I'm at Atlas now" doesn't delete the old fact. It supersedes it. Ask "where did I used to work?" and the agent still knows.

04 / SLEEP

NREM + REM consolidation

NREM strengthens rehearsed memories and lets weak ones fade. REM finds patterns: "Tuesday-morning runs are recurring." Both phases run during the user's sleep window.

05 / WAKE SUMMARY

What changed overnight

The next time the user shows up, your agent surfaces a short narrative of what it consolidated and what patterns it noticed. The most-shared SCM moment.

06 / TENANCY

Multi-tenant safe

Every user_id is namespaced under your account server-side. Cross-account reads are impossible — even if a customer guesses another customer's user_id.

Drop in anywhere.

One Python SDK. One HTTP API. One MCP server. Works with LangChain, OpenAI Agents, Claude Desktop, Cursor, or anything that can issue an HTTP call.

Each request goes to your account's /v1/memories endpoint with a Bearer key. Rate limits are per-key. Tenancy is enforced server-side.

Get an API key Full LangChain guide →
# pip install scm-memory from scm import SCMClient scm = SCMClient( api_key="scm_live_...", base_url="https://scm.run/v1", ) # Store a memory scm.add_memory(text="I run every Tuesday morning along the river.") # Retrieve relevant memories result = scm.search_memory(query="when do I run?") print(result["memory_context"]) # → "I run every Tuesday morning along the river."
What it's for

Built for agents that compound.

Anywhere your AI is supposed to know more about the user this week than it did last week.

Personal assistants

Remembers what the user said weeks ago. Surfaces patterns ("you've mentioned this three times now"). Wakes up with a summary of what it figured out.

Customer support agents

One ticket today, another in three weeks — same customer, same context. SCM consolidates the relationship, not just the conversation.

Coding & research copilots

Knows what you're building, what you tried last Tuesday, what you decided not to do. Doesn't suggest the same dead-end twice.

Research · 35 pages

Sleep-Consolidated Memory: a brain-inspired architecture for lifelong AI agents.

The full architecture, the brutal-test methodology, the workload-sensitivity benchmark, and the four real bugs we caught in the process. Honestly written, including the failure modes.

5 figures 8 tables 322 regression tests 16/16 brutal scenarios
Read the paper
Pricing

Free during beta. BYOK forever.

You bring your own LLM key. We never proxy or charge for inference. We charge for the memory infrastructure — eventually. Right now, all of it is free.

Hobby
Hack on a side project, build a prototype.
$0/forever
No credit card. No expiry.
  • 10K memories
  • 50K operations / month
  • 1 API key
  • Heuristic extraction
  • Community support
Get started
Enterprise
Self-host, SLA, security review.
Custom
Talk to us
  • Dedicated infra
  • VPC / on-prem deploy
  • Postgres + pgvector
  • Custom retention
  • SOC 2 (in progress)
  • Phone + Slack support
Contact us

Give your agent a night.

Sign up, mint an API key in 30 seconds, drop it into your LangChain code. Your agent learns the way humans do.