Choose the work
Email replies, folder reviews, summaries, reports, routines. Start with something you already understand.
Free · Source available, community licensed · Built for Apple Silicon
Of course this is how agents should work.
Powerful AI agents. None of the burden. OOMU runs your everyday work on your own Mac, asks for the cloud only when a task truly needs it, and shows you the plan before it acts.
✓
Find messages that need a reply
✓
Draft replies
3
Review drafts
Nothing sends until you approve.
Review & send
AI can finally do the work.
Read what you don't have time to read. Sort the mess.
Draft the reply. Prepare the report.
For most people, it's still too hard.
OOMU® begins with a different assumption.
You already know your work. The software should adapt to you, not the reverse. OOMU carries the complexity, so what you see stays calm, clear, and ready to use.
How it works
Open OOMU. Then four simple moves.
No command line. No setup maze. No guessing what the agent can touch, or what it will cost to run.
Build the steps
A workflow is a set of clear, visible steps you can see, not a careful prompt you have to perfect.
Review the plan
See exactly what OOMU will read, prepare, and ask before it runs. Change anything before it acts.
Run it
When an action matters, OOMU asks first. When it's done, you can see exactly what happened.
A simple example
It does the work around the decision. It doesn't take the decision.
OOMU removes the tedious work around judgment. It does not remove the judgment.
Take email. Most days disappear into reading messages, deciding what needs a reply, and writing the same responses again and again.
OOMU finds the messages that need attention, drafts the replies, and tells you when they're ready.
You review them. You click send.
Why OOMU is different
Most agents are expensive because they think out loud with your API key.
You should pay for the cloud when you actually need the cloud. Not on every keystroke.
Other tools send every small step to a paid cloud model, so a one-line rewrite costs the same kind of money as a deep analysis. Spend climbs with how often you press Enter, not with how hard the work actually is.
OOMU is built the other way around. Routine work runs on your own Mac, for free and in private. It reaches for a frontier model only when a task genuinely earns it, and sends a focused, pre-trimmed request when it does.
The big AI companies have little reason to fix this; their business runs on token consumption. OOMU is built to reduce it.
Built for Apple Silicon
A deliberate choice, not a limitation.
We chose the Mac because it is the best place to run local intelligence you can trust.
Local AI needs more than a fast chip. It needs room to hold the model, your context, and your work at the same time. Apple Silicon's unified memory gives the whole system one fast pool to draw on, so a capable model runs inside an ordinary Mac. No AI workstation to assemble, no external GPU, no server to rent.
That is why the Mac is part of how OOMU is built, not just where it happens to run. Because the platform is one coherent system, OOMU arrives as finished software: native, signed, and ready in minutes, with no terminal to set up.
We intend to stay with Apple Silicon for as long as it remains the best ground for this work.
Trust, built in
Trust isn't a setting. It's how the product works.
Not a warning screen. Not a paragraph in a policy. It's the result of hundreds of decisions made in your interest.
Visible workflows
Build from clear steps you can see, not careful prompts and hope.
Human review
When an action matters, OOMU asks first. You stay in charge.
Local where possible
Routine work runs on your own Mac. Private work stays close.
Clear records
When work is finished, you can see exactly what happened.
Local where possible
Routine work should stay close to home.
Your default intelligence runs on the machine you already own. Routine work is handled on your Mac, so it never has to leave it, and the cloud is reached only when a task truly calls for it.
Privacy as a property of the architecture, not a promise to be trusted.
Multiply the power
Give your agent a real, new skill. In one file.
A Mod is not a clever prompt or a personality. It's an actual capability you drop into OOMU, the way you drag an app into your Applications folder. One file, fully contained, and it teaches your assistant to do real work at an expert level.
Install a research Mod and OOMU can read and reason across thousands of pages of your own documents, locally, with citations. Add a workflow Mod and it quietly runs your repeated business processes in the background. Stack two together and you've built a bespoke partner shaped to exactly how you work.
You're no longer teaching a chatbot who you are. You stack the capabilities you need, and OOMU executes.
Built for real work
For people who have work to do.
Doctors. Attorneys. Consultants. Researchers. Small business owners. They don't need to become AI experts. They're already experts in their own work, and OOMU respects that.
Two editions, one experience
Individuals need a great app. Organizations need a great app with control.
Community Edition
The real product, open to everyone.
Source available and community licensed, built for individuals, builders, researchers, and small teams. The useful parts are never locked away.
Enterprise
The same app, with the controls teams need.
Deployment, access, usage visibility, security, and team controls, without changing the heart of the product.
Learn AI without the fog
AI shouldn't feel like a private language.
The Learn section explains the basics in plain English: generative AI, agents, workflows, local and cloud models, and the terms people keep hearing but rarely see explained.
You don't need to become an AI expert to use OOMU. But the more you understand, the better your decisions.
Ready to go further? The documentation walks through installing OOMU, building assistants and capabilities, and how it all works under the hood.
Get notified
Be the first to know when OOMU launches.
OOMU is launching soon. Leave your email and we'll send you a single message the day it's ready to download.
We only use your email to tell you when OOMU launches: one message, then we stop. We will never sell it, share it, or use it for anything else. See our Privacy Policy.
Powerful AI, without the worry.
OOMU
Trust, built in.