myAI
Continuity for AI agents — memory that survives the session

You go to sleep, you wake up, you are still you.

Everyone is racing for a smarter model — but the smartest employee in the world is useless with amnesia every night. Your AI agents don’t remember you. myAI fixes that: memory, state, tasks, and budget that survive the session. It’s not a better brain — it’s the ability to keep being someone.

Memory
A git-versioned brain — sessions are commits, wrap-ups are merges.
State
Project state and handoffs every agent boots from. No cold starts.
Tasks
A queue that survives the session — a runner works it off-hours.
Budget
Token metering that persists, so agents pace themselves across days.
Self-hosted in five minutes — Docker + Node 20
$ npm i -g ai-management && myai init . && myai up
Start guided setup — 4 steps →signup → connect a repo → queue a task → watch it ship

Start free in 30 seconds

No credit card. Sign in with your email after.

Already have an account? Sign in

The 15-second proof · demo coming with launch
1
A session dies mid-task. Context gone.
2
Open a different agent — any MCP-capable one.
3
It greets you with your project state and picks up the task. No prompt.

Not a demo — these are live numbers from the platform running this page

64
specialist agents
137
repeatable skills
35
repos under management
22608
tasks shipped to review
See full capabilities →Watch the live metrics →

Three commands to an agent that knows you

No SaaS onboarding, no data upload. Install, point it at a repo, and the next agent you open boots with your full context.

1Install once
$ npm i -g ai-management

One global package. Docker + Node 20, nothing leaves your machine.

2Point it at a repo
$ myai init .

Scaffolds memory, state, handoff and a task queue into the project. `myai up` starts the local dashboard + off-hours runner.

3Your agent boots knowing you
$ # open any MCP agent

Every session starts from your saved state — project, decisions, what’s next. No re-teaching, no cold start. Ever.

Continuity is what makes the loop possible

Because state survives the session, work can span sessions: idea → app → scheduled autonomous work → mobile ops → customer support. Agents that forget can’t do this.

✦
1.Idea
Describe an app or a task in plain English.
⊞
2.App
Agents scaffold and build it — full stack.
◷
3.Schedule
Queued work runs autonomously, overnight.
◉
4.Ops
Approve from your phone with “ship it”.
✉
5.Support
Tickets triage straight back into the build queue.

Not another memory SDK

Most tools people compare us to solve one organ of the problem for a different buyer — and they’re good at it. myAI is the whole continuity loop for the operator running their own repos. Honestly: if you’re building one of the things below, use that instead. Often the right answer is both.

Mem0

A memory SDK you embed in an AI product for your app’s users.

Pick it when
Shipping an app that needs per-user memory.
Zep

A temporal knowledge-graph memory service — tracks when each fact was true.

Pick it when
Building an agent over data that changes over time.
Letta

Memory-as-OS (MemGPT) — an agent that self-manages RAM/disk memory tiers.

Pick it when
Researching self-editing long-context agents.
Pick myAI when

You’re losing an hour a day re-teaching agents who you are — across your own repos, on a Claude subscription. Memory plus state, handoff, a task queue with an off-hours runner, budget, and a fleet console. Self-hosted, plugged into any MCP-capable agent.

Read the full, honest comparison — compare.md →

The whole stack is free. Pay when continuity should follow you.

Open core, bring your own key — we never meter your tokens. Upgrades check out with Stripe inside the dashboard.

Solo
Freeforever · open source

The full self-hosted stack. This is the product, not a trial.

  • ✓Persistent memory, state, tasks & budget
  • ✓Dashboard + off-hours runner
  • ✓MCP hub — bring any agent
  • ✓1 operator tenant, your keys, your machine
Install free
Pro
most popular
$19per month

For anyone with two devices — continuity that follows you.

  • ✓Cross-machine sync (managed Atlas)
  • ✓Hosted dashboard access
  • ✓Continuity metrics history
  • ✓Priority support + early features
Start free, upgrade in-app
Team
$49per user / month

Shared memory for AI-native teams — everyone’s agents, one brain.

  • ✓Shared tenant memory
  • ✓Invites + role-based access
  • ✓Per-member budgets
  • ✓Team fleet console + audit log
Start free, upgrade in-app

Enterprise — SSO/SAML, compliance exports, air-gap install support, SLA — is custom. Talk to us.

The data-locality guarantee

Read the full guarantee — SECURITY.md →
Local by default
Memory, state, tasks and embeddings live on your machine. Set nothing and nothing leaves the box.
Your keys, your model
BYOK — your Anthropic subscription or key, or fully-local Ollama. We never proxy your traffic.
No token metering, no data sales
We don’t meter your tokens and we will never sell your data. Exports are secret-scanned before they leave.

Wake up to an agent that still remembers you.

Install free, connect a repo, queue a task — and tomorrow, every agent you open picks up exactly where you left off.

Start guided setup — free →