Say you have fifty PDFs to turn into assistants. Docutrain will not swallow all fifty in one gulp — three documents can be in flight at a time, and they train one after another. But the queue is built so each round of three takes about thirty seconds of your attention and then runs unattended, which is the part that matters when you are working through a pile. Queuing applies to files only; the Text, Audio and Web tabs upload one item at a time.
Building the queue
There are three ways to start one, all on the PDF / DOC tab of the Train New Assistant window: drop several files onto the PDF / Documents card at once, select several in the file picker, or select one file and then click Add another document beneath it. The line under that button reads "Queue up to 3 documents — they train one at a time".
The window keeps its single-document layout for one file and switches to the queue view as soon as a second arrives. Remove files until only one is left and it switches back, keeping the Document / Slide deck choice you already made, so you are not asked twice.
The same file cannot be added twice — a duplicate is skipped with "handbook.pdf" has already been added. If a drop contains more files than you have slots for, the extras are skipped with "You can queue up to 3 documents at a time."
The queue view
Three parts, top to bottom.
Defaults for all holds an Access menu (Public, Registered users, Group members, Admins only), a Category picker, and the Auto-generate cover image switch. Every file you add inherits these, and changing a default updates every file you have not set individually.
The queue row shows up to three cards side by side, headed "Queue · 2 of 3". Each carries the document's title and chips summarizing its settings: Document or Slides, and its access level (Public, Passcode, Registered, Group or Admins). A chip is highlighted when that file overrides the batch default, which is the fastest way to spot the one file set differently from the rest. The X in a card's corner removes it; an empty slot shows + Add file.
The settings panel appears when you click a card, headed "Configuring · filename". It holds the Assistant name, Category, the Document / Slide deck choice for PDFs, and Access. Edit Category or Access here and the field is marked "· overrides default", so later changes to the batch defaults leave it alone.
Two details catch people out. The Access menu in Defaults for all does not offer Passcode, because there is nowhere in that strip to type one; choose Passcode in an individual file's settings panel, where the passcode field appears with it. And token-link access is not offered before upload at all — set it afterwards in the document's configuration, as described in access levels explained.
Submitting, and what happens next
The footer button counts what you have queued — Upload Document for one file, Upload 3 Documents for three. Until at least one file is added it stays disabled with "Add at least one file to continue".
Two checks run before anything is sent. A file set to Passcode access with no passcode blocks submission: Enter a passcode for "Staff Handbook", or choose a different access method. And your plan's remaining document allowance is taken into account, including anything already in flight, so you can never queue more than your plan can hold. If your plan has one slot left, the queue offers one slot — worth knowing before you sit down with fifty files. See plans and limits.
Everything you set is stored with each upload and applied the moment that document is created, so there is no cleanup pass afterwards.
While the queue runs, each document appears in the Processing panel. The one being trained is listed first, followed by those still waiting, each marked Queued. Only one trains at a time; when it finishes, fails or is cancelled, the next starts on its own. If the server restarts or a hand-off is lost, a background sweep picks the queue back up, so documents left waiting are started rather than stranded.
You do not get a completion window per file, which would be tedious across a batch. Once the whole queue has drained, a single Training complete window lists what finished — "3 assistants finished training. Your settings were applied automatically." — with the titles and a note that you can adjust any of them from the documents table.
Email arrives separately, one message per document: "Document ready: [title]" with the page count and share link, or "Processing failed: [title]" with the error and a View document & retry button. If a failed job is retried automatically, you are emailed only for the first and final attempts.
Cancelling and recovering
Cancel on a processing card opens a confirmation that spells out the consequence: "Processing will stop at the next checkpoint. The document will be marked as failed. You can retry processing or delete the document." Documents still marked Queued cancel the same way, and cancelling one frees its slot so the next starts immediately — useful when you realise halfway through that the third file was the wrong version.
Some recovery is automatic. If the server is at capacity when your upload tries to start, you see "Server busy. Retrying in 30 seconds... (attempt 1/4)", with delays increasing to 60, 120 and 240 seconds. Keep the page open and processing begins as soon as capacity frees up. A job that fails partway may also be retried in the background before a final failure is reported.
The rest is manual. A document ending in Error turns its card into a "Processing failed" panel with the name, the error, the file size, and the upload and failure times. Retry runs the whole pipeline again, though it is not offered when the error says the document is too large — split the file instead. Delete removes the failed upload so it stops counting against your document limit, which matters near the ceiling. Dismiss just hides the card. A card reading Stuck gets a Force Retry button. Common causes are in troubleshooting common issues.
Combining what you loaded
Fifty PDFs sometimes want to be one assistant rather than fifty. The Combine (N) bulk action in the Documents Library merges the content of several trained documents into one new document. Select at least two from the same organization; a cross-organization selection or fewer than two leaves the button disabled with a tooltip explaining why.
The Combine documents window is explicit about what it does. The originals are untouched, and the new document is a snapshot that will not update when the sources are retrained. Give it a name in Name for the combined document, review the source list — documents merge in the table's display order — and select Combine N documents.
The copy runs from your browser, with a progress bar tracking each source ("Copying 2 of 4 — title") and the reminder "Don't close this tab — the copy runs from your browser." Sources with no content yet are skipped and named. Images are copied after the text, then an intro and keywords are generated from the combined content; at that stage you can close the window and generation continues in the background. Each copied passage is prefixed with its source document's title, so citations stay traceable. The new document starts private and consumes a document slot.
Two limits. A combined document cannot be retrained, because it has no source file: to refresh it, combine the sources again and delete the old snapshot. And if a combine never finishes because the tab was closed mid-copy, the half-built document sits in the library with a red, unclickable Incomplete chip — delete it and combine again.