Digital assets
The logo, the screenshot you pasted at midnight, the PDF a client sent. Digital assets keeps each one as a record rather than a loose file: one brand, a note of the door it arrived through, a description you wrote, a status when the file is itself the job, and a running list of everywhere it is used.
Digital assets, close up
Doors in
The library
What pulls from it
What is travelling those lines
This is the same map as the one on the front of Solutions, zoomed into one surface. The doors are on the left, the library in the middle, and what leans on it on the right. Most of what runs here stops at a person, and the map shows where.
What it does
You do not have to remember to file anything. An image pasted into a draft, into a document or onto a board is saved into the library on the way in, under that brand, carrying a note of the door it came through. The place you were working keeps a real link to the record.
An upload has to say which brand it belongs to. If your switcher is on all brands the library leaves the field empty and the upload is refused with a sentence explaining why, instead of quietly filing your work under whichever brand happens to be oldest.
A file carries a status from the same set your writing and your product records use, so a picture in progress and a draft in progress read the same way. One of those statuses does real work: nothing is permanently deleted until it has been marked for removal.
Write what a picture shows on the file itself. That description is what the library searches on alongside the filename, and it is what the image tools read back when you point them at a picture you already have.
The same tag vocabulary the rest of the workspace uses. A tag belongs to one brand, and a tag from one brand cannot be pinned to another brand's file, so a picker never offers you somebody else's words.
Anything that is not a program or a script, up to thirty megabytes each, including the file types other libraries reject by accident. What is refused is refused by name, so you know which file in a batch was the problem, and anything oversize is caught in the browser before the batch is sent so one bad file cannot fail the rest.
Opening or downloading a file goes through a single route that checks you are signed in, checks the file belongs to a brand you can reach, and tells the browser to treat the bytes as data and never as code. A guessed address gets you nothing, and a download keeps the name you uploaded it under.
How much room you are using, what it is made of by kind and by where it came from, your largest files, how much nothing is using, and how much nobody has tagged. Those sit as widgets on your dashboard rather than a report you have to go and build.
Where it came from
Most libraries are a folder you have to remember to put things in, so half your files are in a downloads folder and the other half are attached to one document that nobody else can open. This one fills itself, because the places you already work know how to file. Several doors, one library, and a note on every record saying which door it came through.
You dropped a batch on the library, browsed for it, or pasted a screenshot straight from your clipboard.
What the library keepsFiled under the brand you named. It will not store anything until you have named one.
You pasted or dropped an image into a draft while you were writing it.
What the library keepsA record before the draft is saved, and the draft shows up in the file's own list of what is holding it.
You pasted an image into a task, a note, a decision record or a release note.
What the library keepsThe same library, the same brand rules, and a link the document keeps rather than an address that rots.
You dropped a picture onto a whiteboard mid session.
What the library keepsFiled under that board's brand, with the board written down as one of the things using it.
You set a logo, a light or dark variant, a favicon, a thumbnail or a social image.
What the library keepsIt sits in the library like everything else and gains a refusal of its own against being deleted from here.
You generated a picture, or built one from a layout.
What the library keepsFiled on arrival. One built from a layout keeps the settings that made it, so you can reopen it with everything filled in.
That note is not a label somebody has to remember to set, and it is not a folder you can drag a file out of. It is written into the file's address as it is stored, which is also why two files called screenshot.png never collide and why you can filter a library of thousands down to only the pictures that came off a board.
The refusal
This is the part worth reading twice. Removing the picture that three live pages are showing is the kind of mistake you hear about from somebody else, a week later, and no amount of care prevents it when the library is happy to say yes. Here it is refused, on the server, before anything is removed, in three checks.
First
If the file is a logo, a light or dark variant, a favicon, a thumbnail or your social image, the delete is refused on its own terms and you are sent to the screen that owns it. Nobody removes a favicon by tidying a library.
Then
Pieces of writing, documents, emails, boards and image conversations are all counted. The refusal names the number and the kind, so you know exactly what to go and clear. Clear the last one and it becomes deletable.
Last
A permanent delete is only allowed on a file already carrying the removal status, which is a separate deliberate act. Until then the library disables the button and names the step you are missing rather than letting you hit an error.
Every reference is a row written on both sides at the moment you saved, including the ones worked out from an email's body and a board's contents. A text search for the filename would miss all of those and would also find matches that were never really uses.
When a delete does go through, the stored file and its record are removed as one piece of work. If the stored file cannot be removed, the record is kept too, so you never end up with a list pointing at nothing.
The library shows what is holding a file, and warns you about both missing steps, before you press anything. The number it shows you is the number the server enforces.
None of this is a warning in a dialog that a hurried click gets past. The page tells you first, and the server is what says no.
Where this sits
This is rarely where you start. You are in a draft, on a board, or halfway through a post, and a file has to be somewhere. It is the layer underneath the rest of the workspace, and each of those places has a page you can read.
Set a picture as a draft's cover or drop it into the body. Paste one in and it is saved to the library on the way through, so a published draft is never pointing at something that lives only inside that draft.
Read ContentA file attaches to a task, a note or a decision record, and that document then shows up in the file's own list of what is holding it. Deleting the file is refused while it does.
Read ArtifactsDrop an image on a board and it is filed under that brand, with the board recorded as one of the things using it, worked out from what you saved rather than a list you keep by hand.
Read WhiteboardsA destination receiving your post has to fetch the picture itself and cannot sign in. The publishing run mints a link scoped to one workspace, one file and one expiry, good for half an hour, and resizes the picture on the way out to what that destination accepts.
Read PublishGenerated pictures land in the same library. Point the image tools at one you already have and they read back the description you wrote on it.
Read GraphicsAn agent reads your library, chooses by description, and attaches what it finds to a draft or a document. It cannot add, change, tag or remove anything.
Read Your agentYours to keep
Every one of these we would choose again.
A file something is still using is refused, on the server, with the count and the kind named. Most libraries let you remove the picture three live pages are showing and let you find out from somebody else a week later.
The list shows only brands you can reach, and so does the file's own address, so a guessed URL gets you nothing. Moving a file to another brand is a deliberate edit, never a side effect.
An upload that does not say which brand it is for is refused and told why. Filing a file under the wrong brand is silent, and being asked once beats hunting for it later.
Your agent can read this library all day. There is nothing in its hands that adds a file or removes one, so what disappears is what you chose to remove.
A picture can be converted and resized on the way out to what a destination accepts. The file you stored is not touched, and there is no quiet compression pass.
The note of where a file came from is not a label somebody has to remember to set. It is written as the file is stored, which is also why two files called screenshot.png never collide.
Questions
Keep reading
The surface most files are pasted into on their way in.
What happens to a picture on the way out to a destination.
Where a good half of a working library comes from.
Exactly what an agent is allowed to do with your files.
Open the workspace, do one real piece of work in it, and the joins will show you the rest.