Runs & Executions

Every time you run a workflow, 60 Minute Creator keeps a record of it. This page shows you how to watch a run happen live, look back at past runs, cancel a run, re-run just part of your workflow, and check exactly what each node produced.

Running a workflow

On your workflow's canvas, click the "Run workflow" button in the top right action bar.
If your workflow's last run failed, you'll see a "Previous run failed" dialog first. Choose:
- "Cancel" to stop and do nothing
- "Rerun Full Workflow" to start completely fresh
- "Continue from Failed" to pick up from the node that failed, keeping the results you already have
If your canvas already shows results you could build on, you'll see "Use your existing results?" first. See Reusing results you already have.
Otherwise the run starts right away.

Reusing results you already have

Say you made four images, and each one feeds a video. You like the images and want new videos from exactly those images. Click "Run workflow" and you'll be asked:

"Use Existing Images" keeps your images and makes new videos from them. The dialog tells you what it keeps and what it makes, for example "Keep your 4 images and generate 4 new videos from them".
"Regenerate Everything" starts completely fresh.
"Cancel" does nothing.

This works on your latest results and on any past run you open from your run history.

Good to know:

The last AI step in your workflow is what gets made again, together with everything after it. Everything before it is kept. In Image, then Video, then Compose, you get new videos and a new Compose from your kept images.
You're only asked when there's something to keep. A workflow with a single AI step just runs.
If you've pinned a result on any node, you won't be asked. Your pins decide what's kept.
A result is only reused if the node's inputs are still connected the same way. Adding new nodes after it is fine.
Reused results aren't generated again, so they add nothing to that run's cost. The run still counts as a normal workflow run.

Watching a run live

While your workflow runs, the canvas updates in real time so you can see exactly what's happening:

The "Run workflow" button turns into a red "Stop run" button.
When you start from a node's own Run button ("Run only this node" or "Run this node and after"), that node shows it's working the moment you click.
Each node gets a glowing outline as it works: a pulsing orange and red outline while it's running, a green glow once it finishes successfully, or a red glow if it fails. Nodes that haven't started yet have no outline.
The connecting lines between nodes light up too, showing you which parts of the workflow have finished and which are still going.
As soon as a node produces something, like a generated image, you'll see a small preview right on the node itself.
If something goes wrong, an "Error" badge appears in the action bar. Click it any time to reopen the error details.

If a run ever seems stuck and doesn't start within a minute, you'll see a message telling you it didn't start in time.

A workflow runs one run at a time. If you start a run while another run of the same workflow is still going, even one the AI assistant just started, you'll see "Already running" with "A run of this workflow is already in progress. You can follow it on the canvas." right away. Your canvas stays as it was and follows the run that's going.

When you reload the page, the canvas shows the results of the last run that actually ran, and picks up a run that's still going.

Once a run finishes, the glowing outlines fade back to normal after a few seconds.

If any Export node in your workflow was set to download its output, the finished files download automatically in your browser as soon as that run completes.

Stopping a run

Click the red "Stop run" button to cancel a run that's in progress. The run is marked "Cancelled" and its row shows how long it had been running. If the cancel doesn't go through for some reason, you'll see a "Cancel failed" message.

Stop works from the moment you click Run. A run you stop before it has started never starts, so nothing is made and nothing is charged.

Nothing new starts after you stop a run. A video, image or voice that Atlas Cloud was already making can't be stopped there and is still billed, so it keeps going. Its node stays on your canvas with "Finishing…" under it, the bar at the top says why, and the result shows up on the node when it's done, so you don't need to run it again. Until then the run is still going: the Stop button stays greyed with "Stopping…" and you can't start another run from that canvas. A node the stop caught before it was sent says "Stopped, not charged". Nodes after them in your workflow don't run.

The same applies when one node fails: other nodes that are already making something finish and keep their results, and nothing after them starts.

Checking your run history

Click the clock icon in the action bar, labeled "Executions," to open your run history.

At the top, you'll see a summary: your success rate over the last 30 days and your total number of runs. The success rate is shown in green when things are going well, orange when it's mixed, and red when most runs are failing.

Below that is a list of your past runs. Each row shows:

The date and time it ran
Whether it was triggered manually or on a schedule ("Manual" or "Scheduled")
Its status (see below)
The cost, broken down by provider, with a bold total. If the provider hasn't reported a charge in that run yet, the row says so instead of showing an amount.
How long the run took

If a run failed, you'll see a short error summary right under its row. Scroll down and click "Load more" to see older runs. New runs and status changes appear in this list automatically, without needing to refresh the page.

If you haven't run this workflow yet, you'll see "No executions yet, Run your workflow to see execution history here."

Run statuses

A run can show one of these statuses:

Running: still in progress
Completed: finished successfully
Failed: something went wrong
Cancelled: you stopped it before it finished
Skipped: didn't run because another run of this workflow was already in progress. A skipped run has nothing to show, so you can't open it on the canvas

Looking back at a past run

Click any completed, failed, or cancelled run in the Executions list to view it. This closes the dialog and paints that run's results directly onto your canvas, a read-only look back that doesn't change or re-run anything. Skipped runs have nothing to show, so they can't be opened.

While you're viewing a past run, a banner appears at the top of the canvas showing an eye icon, "Viewing run," the date it ran, and its status. If you've since changed your workflow so that some nodes from that run no longer exist, the banner will let you know.

If the run you're viewing failed, the banner shows a "Resume" button that continues that specific run from where it stopped.
Click "Exit" to leave this view and return to your live canvas.

Starting a new run always takes you out of this view automatically.

You can't select a run that's still in progress. If a new run is currently active, none of the rows in the list can be selected until it finishes.

Running just part of your workflow

Every node has its own "Run" button, with a small dropdown arrow next to it. This lets you re-run a single step instead of the whole workflow.

Clicking the main part of the button is smart about what it does:

On a node with no inputs, it runs just that node, or that node plus everything after it if you've run it before.
On a node whose inputs already have results, it runs that node and everything downstream of it.
On a node whose inputs haven't been generated yet, you'll be asked whether to run your full workflow or run everything up to that node instead.

Click the dropdown arrow for more control. You'll see four options:

Run only this node: re-runs just this node, leaving everything downstream untouched (only available once this node's inputs already have results)
Run this node and after: runs this node and everything that comes after it
Run workflow until here: runs everything up to and including this node
Run full workflow: runs the entire workflow from the start

After you run only a single node, any nodes further downstream that weren't touched are dimmed on the canvas, so you can tell at a glance which results might now be out of date.

Pinning the result you want

Every time a node makes something new, its earlier results stay with it. Step through them, find the one you like, and pin it. A pinned result stays: every workflow run you start keeps it instead of making a new one.

Click an Image Generation, Video Generation, Voice Generation, Lipsync or LLM node. Next to its name you see two arrows and a count, for example "2/3": this node has made 3 results and you're looking at the second.
Use the arrows to step through its results. The node shows each one as you go.
Click the pin next to the arrows. It turns orange: this result is pinned.
Click "Run workflow". The pinned result is kept, and the nodes after it are made again from it.

Good to know:

Only a pinned result is kept. Looking at an earlier result with the arrows changes nothing on its own: without a pin, the next run makes a new result, as usual.
You can pin any result, the newest one too.
A pinned node shows its pin even when it isn't selected, so you can see at a glance what the next run keeps.
You can still step through a pinned node's results. The pin stays on the one you pinned, and it's orange only while that result is on screen. To pin a different one, go to it and click the pin. To unpin, go to the pinned result and click the pin again.
When you click away from the node, or start a run, the node goes back to showing what a run uses: its pinned result, or its latest one.
If the node's last run made nothing (for example the provider blocked it), the count shows "–" instead of a number, like "–/3". Its earlier results are still there: step back to see them, and pin one to use it.
This also works when you start a run from your Workflows page, or when your AI assistant runs the workflow for you. Before it runs, your AI assistant shows which nodes are kept.
Every node can have a result from a different run pinned: the image from one run, the voiceover from another.
Using the node's own "Run" button makes a new result for it and takes the pin off. If that run fails, the pin stays on the result you had.
If you rewire what goes into the node, its pinned result no longer fits, so the next run makes a new one.
The Agent node has a pin too, without arrows: it keeps the prompts shown on its card. If you connect another node to a pinned agent, it writes only the missing prompt and keeps the rest. See Pinning its prompts.
Pinned results aren't made again, so they add nothing to that run's cost.
The arrows go back through your 50 most recent runs of the workflow. If a schedule runs the workflow 50 times after you pinned, the pinned result drops out of the arrows and the node shows its latest result. The next run you start still keeps the pinned one.
Scheduled runs ignore pins: every node runs again each time the schedule fires.
On a narrow node, the arrows take the place of its name while it's selected. The small icon stays.

Checking a node's input and output

Every node has an "I/O Inspector" button, alongside Settings, Rename, and Delete, that appears when you hover over it. Click it to open a side panel with the full details of what went into that node and what came out.

The panel header shows the node's status (Completed, Failed, or Running) and how long it took. If the node hasn't run yet, it says "No run data."

Inside, you'll find two sections:

Input: the model used, your prompt, and a summary of anything that came from earlier nodes
Output: the result, whether that's generated text, an image, a playable video or audio clip, token usage, exported files, or an error message if the node failed

Seeing what's running from your workflows list

On your Workflows page, any workflow that's currently running shows a live "Running" badge with a glowing border right on its card, so you can tell at a glance which workflows are active without opening them.