Test and run

Once your steps are in place, you test them one at a time, run the workflow on a few rows, and then set it live so it runs by itself. This page covers each stage, plus where to see results and what everything costs. Where: open a workflow · https://connie.ai/<workspace-id>/workflows/<workflow-id>. Tests are in each step’s panel; running and going live are in the top-right of the top bar; results are on the Runs tab.

​

Test a step

A test runs one step so you can see its real output before you run the whole workflow. It also captures sample data that later steps’ variables use. What a test uses

• Row data comes from the newest record on the Records tab, unless you choose another one.
• Earlier steps’ data comes from each earlier step’s last test, not from a real run. If an earlier step hasn’t been tested, you type a value instead.
• Free steps that only read data may run straight away without a confirmation.
• A test refuses to send if a required field still contains an unfilled variable.

If the cost can’t be worked out in advance, the confirmation says “This action could not be priced before running.” and “Running it may still charge credits. Continue only if you expect that.” Some actions also note “A provider used by this action may bill separately.” Test a trigger: see Test a trigger.

​

Pin output

Pin output tells a step to stop running and reuse its last test result instead. It’s handy while building: pin an expensive enrichment step once, then test the steps after it as many times as you like without paying again.

1. Test the step first. Until you do, Pin output is greyed out: “Test this node first to capture an output you can pin”.
2. Step ⋯ → Pin output. The card shows a Pinned badge: “Pinned — the stored test output will be returned instead of executing this node”.

A pinned step isn’t executed, so it doesn’t charge credits.

​

Run a workflow

The Run button in the top bar changes depending on the trigger: If Run is greyed out, hover it: “Please configure the trigger node” or “Complete all required fields to go live”. A red Missing field pill in its place means a step still has an empty required field; click it to jump there. What “run once” uses

• Event triggers (email, LinkedIn message, webhook and similar) re-use the captured test event you chose with Use as test event for Run Once. If none has been captured, you’ll see “No test event captured yet”. Capture one first: see Test a trigger.
• Schedule creates a new row in Runs.
• Manually creates a new row each time.

After a run you started, the canvas cards show Completed, Error or Skipped, and a toast reads “Flow finished” or “Flow failed — open run to see why”. These canvas badges only appear in the browser tab that started the run; the Runs tab always has the full picture.

​

Set a workflow live

Live means the trigger is switched on: events, schedules and new rows start runs automatically. Until then the workflow is a Draft (shown as Idle on the Automations page) and only runs when you start it. Messages you may see when going live

• “Please configure your workflow before running”
• “Cannot set live: flow has no trigger node”
• “Cannot set live: add at least one action node”
• Can't set live — "Find Email" and "Send Email" are missing 2 required fields: open the named steps and fill the fields marked *.
• “Can’t set live — reconnect your account first”: click Reconnect in the message, reconnect the app on the Tools page, then try again.
• Can't run — "Send Email" has an invalid Email. Fix the value or use a {{variable}}.
• “The account this trigger uses belongs to a different workspace.”
• A message from a connected app, for example Calendly requiring a paid plan.

​

Pause (go back to draft)

Hover the green Live pill and click Pause, or open the Running menu → Stop live. If you try to change the structure of a live workflow, the “This workflow is live” window offers Set to draft. While a workflow is live you can still edit settings inside steps, but not add, move or delete steps. See Editing a live workflow.

​

Workflow settings

Click the gear icon (Workflow settings) in the top bar. The Settings window: “Apply settings to schedule your workflow.” It’s not shown for Schedule triggers; their card holds the same settings.

• Rules
  • Stop flow if node fails (on by default): a failed step stops the rest of that run. See When a step fails.
  • Queue triggers outside of window (on by default): events that arrive outside your running hours wait until the window opens. If you turn it off, those events are not run.
• Running hours: 24/7, Mon-Fri (9 to 5) or Custom (turn days on and off, set time ranges, Add new time, Apply to all days). Ranges can’t cross midnight.
• Time zone for the running hours.
• List sync (list-based workflows): for each mirrored list, “When on, records added to this list are mirrored into the flow automatically.”

Running hours apply to event and webhook runs, not to Schedule triggers. If your LinkedIn steps would go over your LinkedIn account’s daily limits with these hours, Connie suggests a schedule; you can accept it or click Keep my settings. Click Done to save.

​

The Runs tab

Runs is a read-only grid of results: one row per run (for list-based workflows, one row per list row) and one column per step. The first columns show the record and when the trigger fired. Each cell is that step’s latest result for that row; running a row again replaces it. See the details: click a cell to open Cell Details, with Input (what went in, with variables filled) and Output. Copy as JSON copies it.

​

Fix and retry a failed step

Click a Failed cell. The Execution Failed panel shows the error, the inputs it used, and these options:

• Fix with AI: Connie suggests a fix.
• Edit configuration: opens the step’s settings.
• Retry → Retry this node (“Re-run only the failed cell”) or Retry this row (“Re-run the failed and empty cells”). Retrying a row runs it from the top but only fills failed, skipped and empty cells; steps that already succeeded aren’t repeated, “so nothing is sent twice”.
• Restart this flow (for some sending-account errors) and Upgrade plan (for credit errors) appear when relevant.

​

Re-run rows

On Records or Runs, tick the rows, then Run → Run selected rows → Run all selected or Run only empty cells. Rows that are already running are skipped: “The flow is already running on these rows. Stop them first if you want to rerun them.”

​

Stop a run

• One run you started: hover Running and click Stop (“Flow stopped”).
• Manual workflow: Running menu → Stop run (“Stopped 2 flow runs”).
• Specific rows: tick them → Running menu → Stop selected rows (N).
• Everything, and stop automatic runs: Stop live / Pause.

​

Credits

Credits are Connie’s usage currency. See Credits for balances and plans.

• Each paid step charges its price every time it runs, once per row. Example: a workflow with a 1-credit step and a 2-credit step, run on 100 rows, costs up to 300 credits.
• Free: Router, For each, Delay, the LinkedIn checks, Formula, data helpers and Records actions.
• 1 credit per run for common steps such as Send Email, most LinkedIn sends, Api Call, Scrape Web and Search Internet. Enrichment varies by action. AI image, audio and video steps cost from about 100 to 1,000 credits.
• Run AI agent (the AI Agent step): a 1-credit step fee plus AI usage plus any paid tools it uses. That’s why the picker shows “From N” rather than a fixed price. AI Analyzer and AI-written fields are billed by AI usage.
• Inside a For each, paid steps charge once per item.
• Tests charge just like real runs. Pinned steps don’t charge.
• Failed steps are refunded. Credits are set aside before a step runs and returned if it fails.
• Run only on empty cells avoids paying again for cells that already have a result.

Before a run, Connie checks your balance. At zero or below, nothing runs. If your balance is above zero but below the total price of the workflow’s steps, you’ll see the Insufficient credits window: “This flow requires N credits but your current balance is M.”, with Current balance, Required, Shortfall and Tools in this flow. This check counts each step once and leaves out AI usage, so a long loop or an AI Agent can still run out partway. When that happens, the cell reads “Insufficient credits to run this tool. Please top up and retry.” and the run stops. Owners and admins can see credits spent per record and in the usage log. See Credits.

​

Good to know

• Who can run: testing, running, setting live, retrying and stopping all need Can edit access to the workflow. View-only users see “You are in view-only mode”.
• Failure alerts go to workspace owners and admins, not to the workflow’s creator unless they’re an admin: a bell notification and an email such as “Your flow ‘Lead enrichment’ hit an error”, at most once an hour per workflow. Running out of credits and stopping a run don’t send alerts. Change this under Workflow run failures in Notifications.
• Large runs process rows in batches of about 20 at a time.
• No run limit per workspace, apart from 500 selected rows per run and your credits.
• How long results are kept: run results stay as long as their record exists. Deleting a record removes its results.
• The Runs count on the Automations page is the total number of runs.

​

Troubleshooting

​

Related

• Build a workflow — steps, saving and the live lock
• Variables — why a value comes out empty
• Logic and control — error handling and approvals
• Credits — balances, usage log and plans
• Workflow troubleshooting — more fixes