Your first week
Set up 2 of 4
A new workspace is an empty room, and an empty room is the hardest thing to judge a tool by. The first week is about getting enough of your real work into it that the room starts answering back - because almost everything WeMachines does well, it does by reading what is already there.
Here is the order we would do it in, and the parts we would leave alone until week two.
Invite people before you tidy anything
The instinct is to arrange the place first: statuses, labels, a docs structure, a channel per project. Resist it for a day. Everything below works better with people in it, and a workspace arranged for an imaginary team usually has to be rearranged for the real one.
An admin invites by email, from Settings → Team or from Invite at the bottom of the people panel beside any chat room. Pending invitations are listed there too, and each one says what became of its email - so an invitation that bounced or was marked as spam looks different from one somebody simply has not opened yet. Everyone signs in passwordlessly: a six-digit code, or Continue with Google. The one thing worth saying out loud to your team is use the address you were invited at; a different address makes a new empty workspace rather than joining yours.
Every new workspace starts on a 14-day trial of the Start-up plan, and seats on it are unlimited, so bring the whole team in week one. It also gets 1,000 AI credits, which is enough to find out whether the AI is any good at your kind of work before anyone has to talk to finance.
Four places, and what each is actually for
Chat is channels, the way Discord is: a workspace starts with one text channel and one voice room, and an admin can add more. Every channel is public to the whole team - there are no private channels anywhere in WeMachines. One-to-one conversations exist and are called whispers, and they are quiet rather than secret: only the person you whispered is notified, but anyone can read one, and so can the AI. If that sounds uncomfortable, it is worth sitting with for a week. Most of what teams hide in DMs is context the rest of the team needed.
Tasks are T-1, T-2, and read as a list or a board. Statuses belong to your team, not to us: add, rename, recolour and reorder them in Settings → Workspace. Topics group related tasks the way a project or an epic would; labels and priorities are for filtering. One thing to tell everyone on day one: finished tasks are cleared 14 days after they land in a Done status, and the small draining ring on the row is the warning. Anything worth keeping goes in a doc.
Docs are P-1, P-2 - pages edited together in real time, with comment threads anchored to the passage they are about rather than stacked down a margin. This is where the things that outlive a sprint go: how you deploy, what you decided and why, the runbook somebody wrote at 2am.
Huddles are voice, video and screen share, started from chat. They are transcribed while mics are on, and afterwards the AI writes the call up on the Transcripts page. Summaries and transcripts are deleted 14 days after the call - same rule as tasks, same answer: move anything that matters into a page or a ticket.
Give the AI something real on day one
The demo instinct is to test it: “what can you do?”. That answer is never impressive. Hand it a job instead - something you were going to do yourself this afternoon. @-mention it in a channel, in a task comment, on a page:
@AI read #general from this week and write a page with what we decided.
@AI go through the backlog and tell me which tasks are actually the same task.
@AI turn this thread into tickets, one per thing that has to happen.
It reads the workspace - chat, tasks, docs, the lot - and it acts as itself, so every task it files and every page it writes is signed by it and visible in the activity feed. The quality of what comes back scales with how much of your work is in the room, which is the real argument for getting the team in first. Talking to the AI goes further into what to hand over and how it answers.
What can wait
Plenty. You do not need a channel per project on day one - one room where everything happens is easier to split later than five quiet ones are to merge. You do not need custom statuses until the default ones annoy someone. You do not need to tune an AI teammate before you have watched the built-in one work, and you certainly do not need to decide about credits: the included ones will last longer than your first week.
Two things are worth doing early, though, because they change everything downstream: bring your backlog over so the workspace is about your real work, and connect your repo so the AI can read the code your team actually argues about.
Take it off the browser tab
There are desktop apps for macOS, Windows and Linux at wemachines.com/download, and on a phone you install WeMachines from the browser - open the app on iPhone or Android and add it to the home screen. On an iPhone that is also the only way push notifications are possible at all; Safari does not offer them to an ordinary tab. A native mobile app is in progress. Until then the installed web app is the mobile app, and nothing is missing from it.
One last thing, and it is the most useful key on the site: ⌘K (Ctrl+K on Windows and Linux) searches everything the team has - tasks, docs, channels, pages, people - from anywhere. Teach it to everyone on the first day and you will save them the sidebar.
Tomas MemberInside the product, Home carries the same path as a Set up section, and a step leaves the list the moment the team has actually done it. It is shared rather than personal - a step one teammate finishes is done for everyone - so it doubles as a way to see how far the team has actually got.