Following a Creation While It Runs

Start any AI creation and Scholarly opens a progress page for it straight away. It is the one place that answers the question you actually have while you wait: is this working, and how much longer?

You never have to sit on it. Close the tab, go somewhere else in the app, shut the laptop — the work carries on and lands in your library when it is done. The page just makes the wait legible if you want to watch.

What the Page Shows

A step ladder. The stages of the job, with the current one marked: queued, reading your sources, preparing the workspace, planning the structure, writing the script, generating, recording the audio, rendering, packaging, finishing. Which steps appear depends on what you are making — a podcast records audio, a research report writes a report.

Elapsed time and a typical range. How long this one has been running, next to how long this kind of creation usually takes. A lecture at four minutes with a "usually 8–14 min" beside it is a lecture that is fine.

Detail under the step. Where a creation has countable parts, the page counts them: chapters recorded for a podcast, scenes planned and scenes recorded for a video lecture, source batches read for a long document. A long generation shows real movement rather than one step that appears stuck.

What you asked for. Your sources, prompt, language, length, and chosen model, so you can check the request was what you meant before the result arrives.

Agent activity. Creations that are built step by step — video lectures, slides, documents, spreadsheets, deep research — show what the assistant is doing right now: whether it is working or waiting on a long step, which model it is using, and which tools it has reached for most recently.

Getting Back to It

From Home — anything still running sits in the In progress group at the top of your Recent list. Click it to reopen its progress page.
From a project — running and failed items in a project open the same page.
When it finishes — the creation opens as soon as it lands, and appears at the top of your library.

When It Is Taking Longer Than Usual

If a creation runs well past its typical range, the page says so plainly: "This is taking longer than usual." Most of the time it still finishes on its own. If you would rather not wait, the same panel offers a way to stop it, where stopping is supported.

Cancelling

Where a creation can be cancelled, Cancel generation appears on its progress page. You are asked to confirm first, because anything generated so far is discarded and the work cannot be resumed — you start a new one instead.

Narration and transcripts can be cancelled while they are still queued.

A cancelled creation is clearly labelled as cancelled rather than as a failure, and it does not offer failure-recovery actions, because nothing went wrong.

Cancelling a slide deck, document, or infographic that had already finished cancelling no longer reports an error, and cancelling a Best-quality generation returns its daily Best slot.

When It Doesn't Finish

The page turns into a plain explanation of what stopped it and offers only the actions that could actually help — Retry, a different model, a narrower page range, or a look at the source that caused it. See When a Creation Fails for every reason and what it means.

Work that fails does not cost you an AI Creation Credit; the credit is returned automatically.

If the connection behind a generation drops part-way, Scholarly recovers and continues from what it had rather than starting over, and the page says it is recovering while that happens.

Background Tasks and Notifications — what keeps running after you close the window, and how you are told it is ready.
When a Creation Fails — every failure reason and its matching fix.
Scheduled Tasks — starting a creation at a chosen day and time instead of now.
Home and Library — where in-progress work appears on Home.
Plans and Limits — your AI Creation Credit allowance.

Was this helpful?