Questions,answered.

A personal assistant for your most complex, tedious computer work. You hand it a job in a sentence, and it finishes it end to end on computers of its own, checks it and asks before anything that matters. You get the finished work back, not text to paste.

Aldo isn’t a chat window. Each job gets computers of its own: desktop Excel and PowerPoint on Windows, a Chrome signed in as you, a GPU, your code. It checks the work before it calls it done. And it runs on the subscription you already pay for, theirs included.

Desktop Excel, PowerPoint and Word on a Windows computer, signed in with your Microsoft 365 plan (Business Premium, E3, E5 or Apps for enterprise). Any website, in a Chrome signed in with your saved logins. Your code on GitHub, GitLab, Bitbucket or Azure DevOps. Blender, rendering on an RTX 4090.

No. Aldo works on computers of its own, so yours can stay shut. It tells you on your phone or desktop when a job is done, or when it needs you.

Yes. Each thread’s computer shows beside the thread: its browser, its desktop, its Windows screen. Take the mouse whenever you want; while you have it, the agent’s input is refused.

What you give it. Each thread works on a computer of its own, with the files you attach and the accounts you connect. Threads never share a machine, and your saved logins are typed in for it, never shown to it.

Paying, booking, cancelling and applying: asked every time. Email and invites wait for your yes, or a rule you made. Above the spending limit you set, it asks before it spends, and every spend is recorded.

It checks its work before it calls a job done: it ties the deck to the model, reruns the macro, clicks through the app. The evidence comes back with the result. You can watch any job and take the mouse, and code goes through a pull request you review.

Files stay on the thread’s own computer, kept whole while it sleeps, and in your projects. Logins and API keys go in the vault, each sealed on its own. Once saved, a value can’t be read back, even in Aldo’s app.

Aldo doesn’t use your work to train AI models. Your agents send work to the AI plans you connect, under those providers’ terms; Muse Code starts in Meta’s Contributor mode, in which Meta may train on it. Aldo’s own models go only to providers that don’t train on what you send. The privacy policy has the details.

No. Every plan includes Aldo’s own models. If you have a Claude, ChatGPT, Grok or Muse Code plan, your agents use it first, and Aldo’s models take over when it runs out.

A thread’s computer uses 30 credits an hour while it works, 90 more with Windows and 310 more with a GPU. Asleep, it uses none. Plus’s 3,000 credits come to about 100 hours of its own computer.

Agents keep working on your own subscriptions, one at a time, on 10% of your plan’s credits more, free, until the month renews. Extra usage starts off; turn it on, and agents keep going as usual up to the limit you set, $20 to start.

Any time, from Usage → Manage billing. Plans renew every month until you cancel, and a canceled plan runs to the end of the month you paid for.

3 on Plus, 6 on Pro, 15 on Max. Threads that are asleep don’t count, so you can have as many as you like.

Aldo for companies is in early access: a seat for each person, admins who see cost and never contents, and work that stays when people leave. Aldo for companies has the details.

No. They see cost and computers: credits, thread titles, state and machines. Never conversations, files or vault values.

Leave your email in the form at the end of this page. When there’s room for you, an invitation comes to that address.