The Help Center and Dhanu School now live at docs.dhanu.ai, readable without signing in and searchable from anywhere.
Help and School used to live inside the app, behind a login. That made them
impossible to find from a search engine and awkward to link to — you could not
send a colleague a page unless they already had an account.
Both now live on this site. Nothing is behind a sign-in, every page has a real
URL you can share, and ⌘K searches the whole thing.
Ask Dhanu answers from these pages and cites the ones it used.
Every page is available as plain Markdown — append .md to any URL, or use
the copy button at the top of the page.
Your School progress still follows your account, so lessons you have already
finished stay ticked off.
Every workspace now has a short URL you can choose, and apps you publish take their address from it by default.
Open Settings → General and pick the name you want. Availability is checked
as you type, and if the one you want is taken you get suggestions you can click
straight into the field.
The workspace URL also becomes the default for anything you
publish: a new app’s address starts from
your workspace name rather than from a string of characters nobody can read.
You can still override it per app.
Changing the URL needs permission to edit workspace settings — everyone else
sees it read-only. Existing workspaces already have a URL, filled in from the
workspace name.
Clicking inside a spreadsheet cell editor no longer drops you out of edit mode, and the staff creation page can no longer scroll its own header off-screen.
Sheets — clicking into an open cell editor to move the caret quietly ended
edit mode and handed focus back to the grid. The next Ctrl+A then selected
every cell in the sheet instead of the text you were editing, and
double-clicking a word threw away what you had typed. Clicking inside the
editor now keeps you in it.
AI staff — the staff creation page could scroll the whole layout past the
top of the window, taking the header with it and leaving no way to scroll
back. Selecting an option no longer moves the page.
When staff draft an email or a message for you to use as-is, it now comes back in its own card instead of buried in the reply.
Asking for a draft used to give you prose that looked exactly like the sentences
around it. Copying it meant selecting the right paragraphs by hand, and any edit
you made happened somewhere else.
A draft now has a boundary. It renders as its own card in the thread, and from
there you can:
Type in it — your edits stay in the card while you work.
Copy it in one click.
Ask for another version without re-explaining the whole request.
Send it on, when the draft is something that can be sent.
This only applies to writing the assistant produced for you to use. A draft you
typed yourself stays as you wrote it.
A "Catch up" pill appears when you have unread notifications, and opens a card you can clear without leaving what you are doing.
Sign in with unread notifications waiting and a Catch up pill appears in the
bottom-right corner with the count on it. It stays out of the way, and it is not
there at all when there is nothing to read.
Open it and you get the unread items in one card. From there you can mark
individual items read, Mark all read, or jump to the thing a notification is
about — including when it lives in a different workspace, which is switched for
you.
The list holds still while you read it. Marking an item read does not make it
disappear from under your cursor; it stays until you close the card.
Ask for a clip and you get one, on the same board as your images — including animating a picture the assistant already made.
Ask a task for a short clip and it renders one onto the Media board, next to
the images. You can also point at an image already on the board and ask for it
to move: the picture becomes the first frame, and the clip continues from there.
A clip you like can be extended from its last frame.
Video takes minutes rather than seconds, so the assistant does not sit and wait
for it. The clip appears on the board straight away and fills in when it is
ready — you can keep talking in the task, or close it and come back.
Images and video share one board on purpose, because animating an image turns
one into the other. Full details in
Video generation.