Command Palette

Search for a command to run...

Governed memory for production AI agents

Give every authorized agent context it can trust. MemoryOS turns relevant conversations, corrections, and events into current, attributable, prompt-ready context — without replacing your databases, transcripts, or models.

memoryos://live · governed-context-001
live
Governed
validity · provenance · consent
Production-ready
quality gates · conflict resolution
Complements
your DB · transcripts · vector store
Tenant isolated
scoped per user · agent · tenant
0-stage
Ingest → Retrieve pipeline
lifecycle
0 SDKs
Python · TS · REST · MCP
developer surface
0 controls
Quality, conflict, provenance, lifecycle, degradation, isolation
production gates
0.0%
Designed for SLA-grade AI workloads
tenant isolated
memory.add()memory.get()tenant-A · customer-123quality gate ✓conflict resolvedprovenance: msg#42consent grantedcontext · ranked · freshlifecycle: activescope: this_agent_onlymemory.add()memory.get()tenant-A · customer-123quality gate ✓conflict resolvedprovenance: msg#42consent grantedcontext · ranked · freshlifecycle: activescope: this_agent_only
The memory problem

Storage is the easy part.
Production memory must decide what is durable, what changed, which source to trust, what the user approved, and what context the next agent actually needs.

Without governed memory
1User tells agent a preference
2Agent responds in this session
3Conversation ends, transcript is stored
4Next session: context lost — user repeats
Result
Context lost. User repeats themselves. Agents contradict each other.
With MemoryOS
1User tells agent a preference
2MemoryOS ingests the signal
3Extract · reconcile · govern
4Retrieve governed context → next call
Result
Current, attributable, authorized context ready for the next model call. Better next interaction — every time.

Durable context loss

A returning user repeats lasting goals, preferences, and unresolved issues because prior state was never carried forward.

Multi-agent drift

Support, onboarding, recommendations, and copilots each build a different picture of the same person.

Conflicting truth

New facts contradict old facts, but raw vector retrieval returns both — and leaves the model to guess.

Where MemoryOS fits

Your agent has storage.
It may not have governed state.

MemoryOS complements the systems you already use. It does not replace complete transcripts, business records, document retrieval, or the tools that perform real actions. Send MemoryOS the relevant signals from those systems, and retrieve compact governed context before the next model call.

your app stays responsible forbusiness data·actions·final responses
Application database / CRMsource of record

Owns business records, transactions, and live operational truth.

Transcript storehistory

Preserves complete conversations and message history.

Vector searchretrieval

Finds semantically similar records and documents.

Files, skills & knowledge basesstatic

Provide static instructions and general knowledge.

MemoryOSgoverned state← this layer

Governs learned state: what changed, what is current, and who may reuse it.

MemoryOS vs. the rest of your stack

Not a vector database.
Not a transcript store. Not your app DB.

Each system owns a different job. MemoryOS is the only one that governs learned state — what changed, what is current, and who may reuse it.

Capability
Application DB
records
Transcript Store
history
Vector Search
retrieval
MemoryOS
governed state
Storage
Owns the data it writes
Complete conversation history
State
Decides what is durable vs. transient
Reconciles conflicts between sources
Tracks what changed over time (versions)
Governance
Provenance & source authority
Quality gates on writes
Tenant, user, agent isolation
Retrieval
Returns prompt-ready, compact context
Ranks by relevance + recency + authority
Control
User-approved memory (Memory Passport)
Graceful degradation on quota limits
swipe to compare
Native capabilityPartial / possible with custom workNot designed for this
Domain schema registry

One governance layer.
Multiple production paths.

Start with the general engine. Add a domain schema when your agent needs product-specific extraction, retrieval, and safety behavior.

General Engine

default

Durable user state for facts, goals, preferences, and procedures. Use the general engine when you want durable user context without choosing an industry schema first.

factsgoalspreferencesprocedures

EdTech Schema

domain

Extraction, retrieval, and safety tuned for learning products. Tracks learner progress, weak topics, and exam context.

learner progressweak topicsexam contextlearning style

Support Schema

domain

Remembers open issues, account context, sentiment risk, communication preferences, and resolution history.

open issuesaccount contextsentimentresolution history

Memory Passport

user-owned

Portable, consent-controlled memory the user can approve, inspect, correct, and revoke.

approveinspectcorrectrevoke
From signals to governed context

Five stages. One governed context layer.
Keep your model, tools, and agent framework — MemoryOS handles the memory lifecycle around them.

  1. 01

    Ingest

    messagescorrectionsevents

    Relevant conversations, corrections, and application events enter MemoryOS as raw signals.

    Signals arrive from agents, transcripts, and your backend. Nothing is committed as memory yet — they are candidates pending decisions.

  2. 02

    Extract

    factspreferencesgoalsprocedures

    Candidate durable state is extracted with supporting evidence.

    MemoryOS pulls facts, preferences, goals, and procedures — each one attached to the source and the evidence that produced it.

  3. 03

    Reconcile

    mergerevisecorrectconflict

    Existing claims, revisions, corrections, and conflicts are resolved.

    When a new fact conflicts with an old one, MemoryOS uses source authority, evidence, and recency to decide — and preserves the version history.

  4. 04

    Govern

    quality gateprovenancetenant scopeconsent

    Validity, provenance, authority, lifecycle, and access policy are enforced.

    Each memory is checked by quality gates, scoped to the right tenant, user, and agent boundary, and tagged with who may reuse it.

  5. 05

    Retrieve

    rankcompactprompt-ready

    Current, authorized context is returned — compact and prompt-ready.

    Relevant memories are ranked, source-aware, and small enough to use without handing the model an unfiltered history.

pipeline · stage 1 / 5
ingest
Ingest·state change

Signals arrive from agents, transcripts, and your backend. Nothing is committed as memory yet — they are candidates pending decisions.

Interactive product tour

From user signal —
to governed context.

Select a common memory scenario to see how MemoryOS identifies durable information, handles conflicts, applies controls, and prepares context for an agent.

memoryos://playgroundinteractive · simulated
Choose a decision
decision path · 5-stage pipelineawaiting run
Ingest

Receive the signal

Extract

LLM extracts memory

Reconcile

Detect conflicts

Govern

Quality + consent

Retrieve

Prompt-ready context

Decision log
Press Run decision to play the guided decision trace.
memory responseawaiting decision
Relevant context, ready for the model
output·system_prompt_addition
// run the decision to populate governed context
retrievedrankedfreshconsented governed
Developers

Integrate in an afternoon.
Keep your model, tools, and agent framework.

Add add() on the way in and get() on the way out. MemoryOS handles the memory lifecycle around your stack — in Python, TypeScript, REST, or as MCP tools.

Python
3.9+
TypeScript
Node 18+
REST
OpenAPI 3.1
MCP
2024-11 spec
$ pip install memoryo-sdk
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
import os
from memoryos import Memory
​
client = Memory(api_key=os.environ["MEMORYOS_API_KEY"])
​
# Ingest a conversation signal
write = client.add(
messages=[{"role": "user", "content": "I prefer concise answers."}],
external_user_id="customer-123",
)
if write.job_id:
client.wait_for_job(write.job_id)
​
# Retrieve governed context before the next model call
result = client.get(
query="How should I answer this user?",
external_user_id="customer-123",
)
​
prompt_addition = result.system_prompt_addition if result.has_context else ""
what runtime looks like

A full memory lifecycle, in one terminal.

Install the SDK, ingest a signal, watch MemoryOS extract, reconcile, govern, and retrieve — then attach the governed context to your next model call. No mockups, no fake chatbot — this is the actual decision path your agents will exercise.

  • Real add() / get() flow shape
  • Per-stage provenance and consent surfaced in the trace
  • Compact context, ready to inject into the prompt
memoryos · pipeline previewpreview
How teams start

From zero to governed context
in one afternoon.

step 01

Install the SDK

Add the Python or TypeScript SDK to your backend. Bring your existing agent, model, and tools — MemoryOS sits beside them.

$ pip install memoryo-sdk
step 02

Wire add() and get()

Call add() on the way in with conversation signals. Call get() before the next model call to retrieve compact, governed context.

$ client.add(...); client.get(...)
step 03

Govern and ship

Enable quality gates, tenant isolation, and Memory Passport. Watch the audit trail. Ship to production with confidence.

$ quality_gates = on · tenant_isolated = true
Production foundations

Reliable after the demo.
The controls teams usually discover only after their first memory prototype reaches real users.

audit trail · memory#mem_8821
last 90s · tenant-A
12:04:21.553ingestsignal received · support-bot · msg#42
12:04:21.560extractcandidate: Prefers concise explanations · conf 8.2
12:04:21.572reconcileno conflict · merges into user state
12:04:21.581governquality gate ✓ · tenant tenant-A · consent granted
12:04:21.590storememory#mem_8821 · provenance: msg#42
12:04:24.012retrievequery: how should I answer this user · 2 items ranked

Quality gates

Block duplicate, low-signal, over-budget, or rate-limited writes before they pollute memory.

if signal.score < 0.35 → reject
if is_duplicate(memory) → skip

Conflict resolution

Use source authority, evidence, recency, and explicit review paths instead of storing contradictions forever.

old: codes in Python
new: switch to TypeScript
→ revise · keep history

Provenance

Track which service wrote a memory, what evidence produced it, and why retrieval trusts it.

writer: support-bot/2.3
evidence: msg#42 · confidence 8.8
retrieval: weighted by source

Lifecycle controls

Reinforce useful memory, archive stale facts, preserve versions, and handle corrections without losing history.

state: active
  → reinforced → active
  → stale     → archived
  → corrected  → version+1

Graceful degradation

Keep your product responsive when quotas or dependencies force a partial memory experience.

if quota.exceeded:
  return empty_ctx
  log partial_mode = true

Tenant isolation

Keep customer memory scoped to the right tenant, user, agent, and permission boundary.

tenant: tenant-A
user:   customer-123
agent:  support-bot
scope:  this_agent_only

You likely need governed context when

  • Multiple agents, services, or channels update context about the same user.
  • Corrections must replace stale state without erasing history.
  • Sources have different authority, or provenance and auditability matter.
  • Context crosses tenant, user, agent, category, or consent boundaries.

You may not need MemoryOS yet when

  • Your product is a short-lived or single-session prototype.
  • All required context already fits reliably in one prompt.
  • You retrieve only static documents or general knowledge.
  • There are no meaningful corrections, conflicts, permissions, or audit requirements.
Architecture

Where MemoryOS sits in your stack.
One govern layer between your agent and your model.

MemoryOS does not replace your databases, transcripts, vector store, or tools. It sits beside them — ingesting signals, governing state, and returning compact context before the next model call. Your application keeps full responsibility for business data, actions, and final responses.

production stack · request flowgoverned path
01App receives user input

Your application handles the request. Existing systems (DB, transcript, tools) stay where they are.

02Agent calls add() / get()

On the way in: add() the conversation signal. Before the model call: get() governed context.

03MemoryOS governs state

Extract · reconcile · govern. Quality gates, tenant isolation, provenance, and consent enforced.

04Model gets compact context

system_prompt_addition attached. Model responds. App owns the final response and any actions.

MemoryOS govern layerYour existing systemsSignal / data flow
Trust center

Safe to put in
production.

MemoryOS is designed for teams who treat memory as infrastructure. Encryption, isolation, auditability, and compliance controls are built in — not bolted on.

SOC 2 Type II
in progress
GDPR
compliant
DPA
available
roadmap
HIPAA-ready
enterprise

Every memory is encrypted at rest with AES-256. Per-tenant keys ensure cross-tenant isolation is enforced cryptographically, not just at the storage layer.

API keys are scoped to a tenant + agent + category boundary. A key issued for support-bot cannot read memories written by copilot unless explicitly granted.

Choose where your memories live. Data stays in the selected region; no cross-region replication without explicit customer consent.

The audit trail records who wrote a memory, what evidence produced it, who retrieved it, and when. Exportable for compliance reviews.

Isolation is enforced at the retrieval layer — a cross-tenant query returns empty, not an error that leaks existence. Per-tenant encryption keys add a second boundary.

Set retention per category. User-initiated deletion propagates to all replicas, backups, and derived indexes within 30 days. Provenance is preserved but anonymized.

99.97% uptime (90d)
status.memoryo.dev
operational
Security review
Need our SOC 2 packet or a DPA?
Contact security →
Memory Passport

Share memory without hiding control.
Approve, inspect, correct, resolve, grant, and revoke — live.

This is a live console, not a mockup. Click a memory to inspect its provenance. Toggle a grant and watch the consent ledger update.

Memory Passport
customer-123
tenant
tenant-A
memories
4
agents
4
grants
3 active
memories · click to inspect4 stored
agent grants · toggle to test3/4
support-bot
all categories
tutor
learning progress only
recommendations
preferences + goals
copilot
facts + procedures
activity · consent ledgerlive
  1. Approved · support-bot2m ago

    Preference: Prefers concise explanations

  2. Corrected · copilot1h ago

    Fact: Codes primarily in TypeScript

  3. Granted · tutor agent3h ago

    Scope: learning progress only

  4. Resolved · conflict5h ago

    Two sources reconciled via authority

  5. Revoked · recommendations agent1d ago

    Access removed by user

Approve
memory before it's stored
Inspect
what each agent sees
Correct
facts that drifted
Resolve
conflicting versions
Grant
scoped agent access
Revoke
access in one click
Built around real workflows

Context shaped for the product it serves.

Customer Support

Pick up where the last ticket ended.

Remember open issues, account context, sentiment risk, communication preferences, and resolution history — across channels and over time.

customer-123 · last 30 days
open issuebilling discrepancy · $240
preferenceprefers email · 9-5 ET
sentimentelevated risk · last 3 tickets
resolutionrefund issued 2025-11-12
Education

Teach the learner you already know.

Carry weak topics, exam context, learning style, language comfort, and progress signals into every session.

learner progress12 memories
Calculus · integration by parts64%
Reading comprehension82%
Spanish · past tense41%
Memory Passport

Share memory without hiding control.

Let users approve categories, inspect grants, correct facts, resolve questions, and revoke agent access — at any time.

approved
34
revoked
2
agents
3
support-botgranted · all categories
tutorgranted · learning only
recommendationsrevoked
Multi-agent systems

One governed picture of the same person.

Support, onboarding, recommendations, and copilots each build a different picture of the same person. MemoryOS reconciles them.

explore use case
Signals

What teams building with
governed memory actually say.

94%
less context repetition for returning users
6wk
of bespoke reconciliation logic replaced
5-stage
lifecycle fully audited per memory
1 click
for users to revoke any agent's access

"We were hand-rolling a memory layer on top of Postgres and pgvector. MemoryOS replaced six weeks of bespoke reconciliation logic with a single add() call."

Engineering Lead
AI Copilot Startup
Series A

"The Memory Passport was the deciding factor. Our users in regulated industries need to see and revoke what agents remember — MemoryOS ships that out of the box."

Head of Product
Enterprise Support Platform
B2B SaaS

"Conflict resolution alone is worth it. Before MemoryOS, contradictory facts would leak into prompts. Now the model gets one governed, source-aware view."

Staff Engineer
Multi-agent infra
Pre-seed
Illustrative testimonials based on the product's design partner conversations. Metrics reflect typical deployment shapes, not specific customer commitments.
Production at a glance

Built for the shape of
real agent workloads.

These are the numbers that matter when memory is in the hot path of every model call. Illustrative — based on design-partner deployment shapes.

memory operations / min · last 40 tickslive
729ops/min+12.4%
p50
42ms
p95
118ms
p99
204ms
0ms
p95 retrieval latency
0.0M
memories under governance
0
active tenants
0.00%
uptime (90d)
Glossary

The vocabulary of
governed memory.

These are the concepts that distinguish a memory layer from a database, a transcript store, or a vector search. Click a term to expand its definition.

Pricing

Pay for governed context.
Not for idle storage.

Start small. Upgrade when your agents reach real users. Every plan includes the full lifecycle — ingest, extract, reconcile, govern, retrieve.

Free

For prototyping and side projects.

₹0/ mo
free forever
Start free
  • 5,000 memory operations / mo
  • 2M memory tokens / mo
  • 3 calls / user / minute
  • Core memory lifecycle
  • Community support
popular

Starter

For early production agents.

₹1,800/ mo
per tenant / mo
Choose Starter
  • 50,000 memory operations / mo
  • 25M memory tokens / mo
  • 10 calls / user / minute
  • Core memory lifecycle
  • Usage alerts

Growth

For growing multi-agent products.

₹6,000/ mo
per tenant / mo
Choose Growth
  • 500,000 memory operations / mo
  • 250M memory tokens / mo
  • 30 calls / user / minute
  • Core memory lifecycle
  • Usage alerts

Scale

For high-volume production systems.

₹18,000/ mo
per tenant / mo
Choose Scale
  • 1M memory operations / mo
  • 500M memory tokens / mo
  • 80 calls / user / minute
  • Core memory lifecycle
  • Priority usage alerts

Enterprise

For regulated, multi-agent systems.

Custom
custom
Talk to us
  • Custom operations & schemas
  • Tenant isolation + SSO/SAML
  • Custom domain schemas
  • Graceful degradation + SLA
  • DPA · SOC 2 · audit exports
  • Dedicated solutions engineer
simple monthly billing secure Razorpay checkout keep your stack SOC 2 ready
FAQ

Questions teams ask
before they ship.

Short, technical answers. If something here doesn't cover your case, talk to us — we work through architecture with every team.

No. Vector search finds semantically similar records and documents — it does not decide what is durable, what changed, or who may reuse it. MemoryOS governs learned state: facts, preferences, goals, and procedures, with provenance, conflict resolution, and access policy. You can still use a vector store alongside MemoryOS; we complement it, not replace it.

Changelog & roadmap

Shipping continuously.
What's live, and what's next.

v1.0Sep 2026shipped

General availability

  • Domain schema registry (General, EdTech, Support)
  • Memory Passport with grant/revoke + provenance drawer
  • MCP server for MCP-compatible agents
  • Quality gates, conflict resolution, tenant isolation
v0.9Aug 2026shipped

Production foundations

  • 5-stage lifecycle (ingest → retrieve)
  • Provenance chain + audit trail
  • Graceful degradation on quota limits
  • Python + TypeScript SDKs, REST API
vnextQ4 2026next

On the roadmap

  • Streaming extraction (SSE) for sub-second traces
  • Self-hosted / on-prem deployment
  • SOC 2 Type II + DPA
  • Custom domain schemas (financial services, healthcare)
start with your real product

Give your next AI interaction
the context the last one earned.

Try the simulated playground, follow the quickstart, or talk with us about production onboarding.

0 min
to first memory
0 SDKs
Python · TS · REST · MCP
0 lock-in
keep your stack
0.0%
tenant isolated · SLA-grade