Live Generation Status on the Prizmad Dashboard

The worst part of generating video ads used to be the waiting — specifically, the stuck waiting. A generation that sits on "processing" for an hour, a refresh that shows the same card, a render that quietly failed while you stared at a spinner. This week we shipped live generation status on the dashboard: every AI Studio generation now updates in real time as it moves through the pipeline.

The change is a fix as much as a feature. Our agent chat starts generations through the same pipeline as the wizard, but a generation only reached a terminal state when something actively polled its status route — and the chat agent stopped polling after a check or two. The result: agent-started generations could stay stuck at "processing" on the dashboard even after the video was actually done. Now the dashboard polls every active generation itself, so the status you see is the status that's true.

What shipped

Three pieces, all live on the dashboard today:

1. Live status polling. The dashboard queries active (pending/processing) generations and runs the canonical status route for each one every 3 seconds while the tab is visible. Polling pauses when you switch tabs and resumes the moment you come back — no serverless calls fire into the void while nobody's watching.
2. Batched, race-safe updates. Active generations are checked in small sequential batches (8 at a time per cycle), so a burst of renders can't fan out into dozens of simultaneous requests. Terminal writes are race-safe, which means the agent and the dashboard can poll the same generation at the same time without corruption.
3. Cards that flip live. When a generation completes, its card updates in place with the result — no page reload. New generations appear and bump the totals, failed renders surface their error immediately, and pending cards now carry a clear Queued label instead of a vague spinner.

The merge logic is idempotent: the same snapshot can be applied repeatedly without duplicating cards, and a single failed status call can't corrupt the feed — the dashboard keeps the last good state and retries on the next cycle.

Why this matters for ecommerce ad teams

Creative testing is a volume game, and volume without visibility is chaos. Here's what live status changes in practice:

• Batch testing becomes monitorable. Queue eight variants at once and you can watch them flip Queued → Processing → Done in sequence instead of guessing when to come back. You'll know within seconds when the batch is ready to review.
• Failures stop hiding. A failed render with no error message costs you a creative slot in your testing cadence. Now failures surface instantly with the error attached, so you can re-queue or change the approach instead of discovering a dead card hours later.
• No more blind refreshing. The dashboard updates itself while you work on hooks or offers in another tab. Come back to a finished batch — or a clear picture of what's still rendering.

The economics don't change: generation still draws standard tokens from your plan (Launch $39/mo, Starter $99/mo, Pro $249/mo), and a finished avatar ad still lands in roughly five minutes from a product URL. What changed is the feedback loop around it — the difference between submitting work into a black box and watching it ship.

Try it

Sign in to the Prizmad dashboard on any paid plan and generate — from the wizard, AI Studio, or agent chat — and watch the status flip live. If you're not on a plan yet, pick one: the same token math covers generation, and now it comes with a progress bar that actually tells the truth.