Safety First

Before anything else — VisionAI Workspace is safe. Our Google OAuth integration has completed Google's app verification process, so you'll see a standard, verified consent screen when you sign in — not an "unverified app" warning.

Here's what actually happens with your data: When you connect your Google account, VisionAI Workspace creates a Google Sheet inside your own Google Drive. That sheet is your workspace and the system of record. We write to it while you're using the platform, keeping only a short-lived server-side cache and account/usage data to run the service.

Signing In

Click "Join the Beta" or "Log In" from the homepage. You'll sign in using your Google account. This is the only login method currently available.

Understanding Google Permissions

VisionAI Workspace requests three Google permissions. Here's exactly why each one is needed:

What is an Orbit?

An Orbit is VisionAI Workspace's core unit of work. Think of it as a project or goal — something you want to build, launch, or create. Each Orbit contains a set of tasks, each handled by a specific AI worker, all tracked in your workspace Sheet.

You can have up to 4 active Orbit pads in the Beta tier — each one a separate project running independently within your workspace. Think of them as dedicated workspaces for each of your goals, all visible from your dashboard at a glance.

Creating an Orbit (The Wizard)

When you create a new Orbit, VisionAI Workspace guides you through the Orbit Wizard. You'll describe your goal in plain language — no special formatting required. VisionAI Workspace's Brain will then:

• Optimize and clarify your input into a structured goal.
• Suggest the tasks needed to achieve it.
• Automatically assign the right AI worker to each task.
• Present a task plan for you to approve, reject, or customize before execution begins.

Tasks & Canvas

Each approved task appears on your Orbit canvas — a visual board showing all your tasks, their assigned AI workers, and their current energy state. Click any task to open it and work directly with your AI worker via the dedicated chat thread.

After you accept an artifact, the platform automatically suggests follow-up tasks and assigns them to the right AI worker. These appear as ghost cards on your canvas — you can confirm or dismiss them.

Energy States

Every task has an energy state that shows where it is in its lifecycle:

You may also see a translucent "Beacon" marker on the canvas — that's not a state a task moves through. It marks an AI-suggested follow-up task preview (a "ghost" card) connected to its parent, before you confirm it into a real task.

What is an AutoLoop?

An AutoLoop is recurring, scheduled AI work inside an Orbit — set it up once, and it runs on its own cadence (daily, weekly, or monthly) without you having to re-trigger it every time. Instead of manually asking an AI worker to redo the same kind of task on a schedule, an AutoLoop does it automatically and surfaces the result for you to review.

You create one from an accepted artifact inside a task — VisionAI Workspace uses that artifact as the starting point for every future run, so each cycle builds on real, approved work rather than starting from scratch.

MomentumLoop vs. InsightLoop

Every AutoLoop is one of two types, chosen when you create it:

Scheduling and cooldown days

Pick daily, weekly, or monthly when you set up the loop. A daily MomentumLoop also follows a 2-on/1-off cooldown pattern — two days of new work, then a rest day — to keep quality up and avoid producing more drafts than you can realistically review. The Refresh Pack can skip an upcoming cooldown day if you need uninterrupted daily output for a stretch.

Auto-pause: a built-in check-in

An AutoLoop is meant to work with you, not run forever unsupervised. If a MomentumLoop piles up 5 runs in a row that you haven't reviewed and accepted, it pauses itself rather than producing a 6th. If an InsightLoop reports "nothing new" for several cycles in a row, it pauses instead of repeating the same quiet update. Either way, you'll get a notification explaining why it paused and a one-tap way to turn it back on whenever you're ready.

Shared workspaces

A workspace has one owner. The owner can invite other people into it by email, and chooses which orbits each person can reach — being in a workspace does not mean seeing everything in it. That access can be widened or narrowed later, and anyone left with no orbits at all is removed from the workspace rather than left with an empty shell. Both changes are announced to the person they affect.

There are two roles, editor and viewer. Everything the workspace produces is saved to the owner's Google Drive rather than split across each member's own, so there is one source of truth for the work.

You can belong to several workspaces — your own, plus any you have been invited to — and switch between them in the app. They stay entirely separate: orbits, tasks, artifacts and conversations never cross from one to another.

Crew messages

People sharing a workspace can message each other without leaving the product. Every orbit has a room containing exactly the people with access to that orbit — access to the orbit is what puts someone in the room, so removing it removes them from the room too. Any two people who share at least one orbit can also message directly.

A message can carry images, audio and video that play inline, files up to 8MB, GIFs, links with a preview, and reactions. It can be edited for 15 minutes after it is sent and not after. Unread conversations show in the notification bell, and a message can raise a push notification on a device that is not currently looking at the app.

Data & Privacy

Task Execution