YOLO Assistant is the guide built into YOLO Studio. It is not a separate app you open or a chatbot in a corner — it is the bar you are already typing in. It knows what you are building, what is running, and what is stuck, and it turns a sentence into work your agents can actually do.
Where to find it
There is no assistant page to visit and no app to install. Sign in, and it is already there.
01
Open your dashboard
The command bar sits at the bottom of the screen. Click it and it opens into a conversation.
02
Or press ⌘K in a workspace
Inside a workspace the same assistant is one keystroke away — ⌘K on Mac, Ctrl+K elsewhere.
03
Just say what you want
Plain sentences. No slash commands to memorize, no prompt formats to learn.
What it actually does
The assistant is a guide and a dispatcher, not a second execution engine. It reasons about your work, then hands the doing to the agents you already use.
Guidance that knows your workspace
It can see what you are building — the repo, the running tiles, the last thing that failed. So "what should I do next?" gets an answer grounded in your actual project instead of a generic tutorial. Blank preview, failing command, half-finished feature: describe it in a sentence and get a next step.
Holds the whole portfolio
Ask "what needs me?" and it answers across every workspace you own, not just the one on screen: decisions waiting, reviews to sign off, what is running right now, what landed while you were away. It knows what each of your workspaces is actually for.
Remembers, and learns your way
Tell it a preference once — how you like PRs described, which stack you default to, what "done" means on your team — and it carries that forward. You can teach it skills in your own words, review everything it has remembered, and make it forget any of it.
Hands off, then reports back
It can start the work: spin up a lane, dispatch an agent, queue a task. Long jobs do not hold the conversation hostage — it tells you when the thing you asked for is done, and where the result is.
It proposes. You decide.
Anything that changes your project, spends money, or leaves the platform is shown to you first and waits for a tap — the assistant never takes those actions on its own initiative.
Proposals are concrete: the exact action, on the exact target, before it runs. Not "I will handle it."
Autonomy is a setting you own. How much the platform is allowed to land without you is yours to set, and the assistant can never raise its own ceiling.
Scoped, and yours to switch off
It only ever reaches your own workspaces, your own runs, and your own memories — the account boundary is enforced by the platform, not by the model.
One master switch turns the assistant off completely: no messages, no suggestions, no actions, on any surface.
Everything it has remembered is visible to you, and anything can be removed on request.
It has a name and a voice — Octavius by default, Octavia if you prefer the alternate — and you can change it per workspace.
The same security posture as the rest of YOLO Studio applies: isolated workspaces, no stored passwords, credentials that expire with your session.
Things people actually type
No syntax. These are complete requests.
Already in the box
YOLO Assistant ships with YOLO Studio — there is nothing to enable and nothing extra to buy. Sign in, click the command bar, and ask it something.
