Meet Fumu in 60 seconds. Watch the intro

Design notes

A task that can wait three days

Real errands involve waiting: for a reply, an approval or a Monday morning. Fumu is built so that waiting never loses work, and a retry never does anything twice.

Fumu dozing with its shades on, waiting, on a burgundy background.

Ask an assistant to “follow up with the supplier and book the earliest delivery slot” and the interesting part is not the email. It is the three days between sending it and getting an answer.

Most AI assistants live in the moment. They are brilliant for the length of a chat, and then the thread is gone. If anything goes wrong in those three days, you are back to doing it yourself, and you may not even know what was already done.

Fumu is built on the assumption that waiting is normal.

Every request is a task that is never lost

When you ask Fumu for something, it becomes a task with its own record: what you asked, the plan, each step taken and each step still to come. Fumu checks off the steps as it goes.

So when a task has to wait (for a reply, for your approval, for next Monday) it simply pauses at the step it reached. When the reply arrives, it picks up from there. If something restarts in between, nothing is forgotten: the task carries on from its last completed step.

Thinking and waiting are kept apart

Inside each step, Fumu can think as hard as it needs to: try an idea, look at the result, try again, hand part of the job to a specialist. Around the steps, the task keeps a steady, step-by-step record of what has actually happened.

That split is what lets it be both clever and dependable. The thinking can wander; the record never does.

Exactly once, even when retried

Retries are where assistants usually go wrong. A hiccup during “send the email” leaves you wondering whether it went out. Fumu gives every action that changes something a unique label, so if a step is retried, the email goes out once, not twice. The same applies to posts, messages and anything that costs money.

What it feels like

From your side, none of this is visible, and that is the point. You ask for something on Thursday, Fumu sends the email and goes quiet (shades on, dozing). On Monday the reply arrives, the task wakes up, and Fumu books the slot and tells you, with a full record of everything it did in between.

Long waits stop being a failure and become just another step.

Fumu with its pink shades pushed up, thinking, on a burgundy background.

Design notes

Why Fumu asks before it acts

An agent that can send email as you is only useful if you trust it. Here is how Fumu earns that trust one capability at a time, instead of asking for all of it on day one.

Fumu celebrating with sparkles around it, on a burgundy background.

Design notes

Fumu grows one skill at a time

Everything Fumu can do is a capability you can add, pause or remove. New ones arrive with no access, and earn trust the same way as everything else.

Fumu peeking over its pink shades with a knowing look, on a burgundy background.

Brand

Reading Fumu's face

Fumu has nine moods and one rule: its face only changes when a task changes state. Here is what each one tells you at a glance.

See Fumu at work.

Watch the 60-second intro, follow a real run from issue to pull request, and ask for a live walkthrough.

See the showcase