Automate multi-step work that runs on a schedule, when data changes, or when a connected tool sends an event.
Use workflows when you want to take work off your plate, such as welcoming new customers, sending summary emails, syncing data between tools, or running multi-step processes that wait, check conditions, and react automatically.When building an app in Base44, workflows run backend work for you automatically. A workflow can start on a schedule, when your data changes, when an app user talks to an in-app agent, or when a connected integration sends an event. You create and manage workflows using the AI chat in your app’s dashboard.
Workflows is available on the Builder plan and above. If you are on the Free plan, upgrade to a Builder plan or higher to create and run workflows.
The Workflows dashboard, where you create and manage workflows.
Important: Workflows replaces automations. Apps created from July 6, 2026 use workflows, while apps created before then may still use automations. Your app has one or the other, not both. Everything you could do with automations, you can do with workflows, plus more:
Multiple steps: Chain several actions together in one flow, instead of a single action.
Conditions: Take different paths depending on your data, so the right steps run in each situation.
Delays: Pause for minutes, hours, or even days between steps, then continue automatically.
Step-by-step visibility: Watch each run in real time and see exactly which step succeeded or failed.
What can I use workflows for?
Workflows handle almost any backend work you want to happen automatically, especially when a task takes more than one step. For example:
Nurture leads: When a new lead signs up, send a welcome email, wait 2 days, then follow up only if they have not replied or booked a call.
Alert on conditions: Check inventory on a schedule and message your admins only when an item drops below its threshold.
React to data changes: When a new product is added, notify your team and create a task to review it before it goes live.
Keep tools in sync: When a Google Calendar event is cancelled, update the matching booking and email the customer.
Send scheduled summaries: Every morning, email yourself a recap of the previous day’s orders and post the total to a Slack channel.
A workflow is made of a trigger and one or more steps. The trigger decides when the workflow starts. The steps decide what happens, in order, each time it runs: a step can run a backend function (code the AI writes for you), wait for a set amount of time, or take a different path based on a condition.When you describe what you want in the AI chat, Base44 builds the workflow for you. It appears in two places: your app’s file tree and the Workflows dashboard. Opening it shows a visual diagram that flows from the trigger through each step. Each run is recorded, so you can review what happened and fix any issues quickly.
A workflow shown as a diagram, from the trigger through each step.
Example: welcoming a new lead. When a lead is created, the workflow sends a welcome email matched to their fitness goal, waits 2 days, then checks whether they booked a call. If they did, it adds the call to your calendar; if not, it sends a follow-up nudge.
There are 4 ways a workflow can start. When describing what you want to the AI chat, it helps to know which trigger fits your use case. Read more about each one below.
Scheduled workflows run at a time you define in your timezone. You can run one time, on a recurring schedule, or on an interval, for example every 30 minutes. Use these for recurring tasks like daily digests and weekly reports, or for a one-time job at a specific date and time.Example prompts:
Send me a daily summary of all new orders to my email every morning at 8:00.
Archive trip records that have been completed for more than 30 days, every Sunday at midnight.
Keep these in mind for scheduled workflows:
Recurring schedules run at most once every 5 minutes, and you can set them to end on a date or after a number of runs.
A schedule that reaches its end, including a one-time schedule after it runs, is archived automatically.
A recurring schedule with no end keeps running until you deactivate or archive it.
Entity workflows run in response to changes in your app’s data, such as a record being created, updated, or deleted. Use these to react to activity in real time.Example prompts:
Send a push notification whenever I add a new WishlistPlace and write any needed backend function for me.
Send the admin an email whenever a new booking is created.
Keep an entity workflow focused:
Add a condition so it only runs when it should, for example only when an order’s status changes to shipped, rather than on every update.
Avoid having a workflow update the same record that triggers it, which can make it run in a loop.
In-app agent workflows run once when an app user starts a new conversation with an in-app agent, not on every message they send. Use these to stay on top of who is reaching out, for example to get notified when a customer starts a chat, add them to your contacts, or send yourself a summary at the end of the day.Example prompts:
Email me whenever someone starts a new conversation with my agent.
Add the person to my Contacts list when they start a chat with my agent.
Connector workflows run when a connected tool sends an event, such as a new email arriving in Gmail or a calendar event being updated in Google Calendar. The tool must be connected first, and it must support workflow triggers, such as Gmail, Google Calendar, Google Drive, Microsoft Outlook, Microsoft OneDrive, and Slack.You can add a condition so a workflow only runs for the events you care about. This matters most for busy tools like Slack, where you might want to run only for messages in a certain channel or from a certain person.Example prompts:
When a new email arrives in Gmail with the subject "Book appointment", parse it and add a new booking to my app.
When a Google Calendar event is cancelled, update the booking in my app and post a message in the #salon-general Slack channel.
If a tool does not support workflow triggers, you can still bring its events in. For example, to act on new Stripe payments, use a webhook or backend function to save each payment as a record in your app, then use an Entity workflow that runs when a new payment record is created. Or, instead of waiting for the tool to notify you, run a scheduled workflow that checks the tool on a regular basis, for example every hour, and handles anything new since the last check.
To create a workflow, click Create workflow in your app dashboard. Base44 opens a Create a new workflow canvas, where you either start from a trigger or pick one of the suggestions based on your app. From there, you describe what you want in the AI chat and it creates the workflow, asking any questions it needs. You keep refining it in the chat until it works the way you need.
Start from a trigger when you already know how the workflow should begin.To create a workflow from a trigger:
Click Dashboard in your app editor.
Click Workflows.
Click Create workflow.
On the canvas, click Add trigger, then choose how the workflow starts: a scheduled trigger (One time, Recurring, or On an interval), a connector, an in-app agent, or an entity.
In the AI chat, describe what the workflow should do, and answer any follow-up questions the AI asks.
The AI creates the workflow and adds it to your Workflows list.
The Create a new workflow canvas, where you add a trigger to start.
Start from a suggestion when you want a ready-made idea tailored to your app.To create a workflow from a suggestion:
Click Dashboard in your app editor.
Click Workflows.
Click Create workflow.
Under Or start from these suggestions, choose a suggestion.
Review the pre-filled prompt in the AI chat, adjust it if needed, and send it.
The AI creates the workflow and adds it to your Workflows list.
Starting a workflow from a suggestion.
Get the best results from the AI:
Describe each step and outcome clearly so they show up in the diagram.
For a condition, give each path a distinct action, otherwise the AI might put the logic inside a single backend function instead of showing a separate path.
If a step needs a new backend function, just mention it and the AI writes it for you.
If your app still uses automations, you can move all of them to workflows in one step. Base44 recreates each automation as a workflow, so you do not have to rebuild anything.
What happens when you switch:
Your automations become workflows: Each automation is recreated as a workflow, keeping its original trigger and schedule.
Your history carries over: Run counts and stats move with each workflow. Past run logs stay with your original automations, which are kept as archived.
Archived automations come across too: Any automations you already archived move over as well and stay archived.
Nothing stops running: Your automations keep running until the switch finishes.
To switch from automations to workflows:
Click Dashboard in your app editor.
Click Automations.
Click Switch to Workflows.
Review what happens when you switch, then click Start now.
After you start using workflows, you can control when they run, change what they do, rename them, trigger them on demand, or clean up workflows you no longer need. Each workflow shows an Active or Inactive status, and you can deactivate one to pause it and reactivate it whenever you need it.
Deactivate a workflow when you want to stop it running, and activate it again when you need it. Deactivating stops new runs from starting, but the run history stays available.To turn a workflow on or off:
Click Dashboard in your app editor.
Click Workflows.
Click the More Actions icon on the workflow.
Click Deactivate or Activate.
You can also open a workflow and use the Active toggle in the details panel.
Deactivating a workflow from the More Actions menu.
To change what a workflow does, edit it from its diagram or just describe the change in the AI chat. Every edit option opens the AI chat with a pre-filled prompt and the AI makes the change for you, so there is nothing to edit by hand.
Editing a step
Click a step’s Edit icon to change what the step does or replace it with a different step.
Editing a step.
Editing the trigger
Click the trigger’s Edit icon to change which events trigger the workflow.
Click the Replace trigger icon to switch to a different trigger, such as a scheduled, connector, in-app agent, or entity trigger.
Replacing a workflow's trigger.
Adding a step
Click the add button below a step, then choose Action or Condition.
Adding a step to a workflow.
Changing a workflow’s steps creates a new version. Runs already in progress finish on the version they started with, and new runs use the latest version.
Run a workflow right away when you want to start it on demand. This helps you test changes, confirm that your configuration works, or handle something urgent between scheduled runs.
Important: Run now executes real actions using your current configuration, such as sending real emails or updating real records. It is a one-time manual run that does not affect your schedule or trigger, and it appears in the run history labeled as a test. When earlier runs exist, you can also replay the workflow using the data from a previous run.
If a workflow is running and you need to stop it, you can cancel the run while it is still in progress. The cancelled run shows as Cancelled in the run history.To cancel a run:
Click Dashboard in your app editor.
Click Workflows.
Click the workflow to open it.
Click the run that is in progress, then click Cancel run.
Every workflow records each run so you can watch it happen in real time and review it afterward. Review a single workflow’s runs from its details panel, or see runs across all your workflows on the Activity tab.
As a workflow runs, each step in the diagram shows a green checkmark when it completes successfully, so you can confirm at a glance that it works and see exactly where it stops if a step fails.
When you open a workflow, the details panel shows its Success rate, Avg. runtime, and a Run history of every run. When something goes wrong, the run shows the exact step that failed and the error, so you know where to look without guessing.To view a workflow’s runs:
Click Dashboard in your app editor.
Click Workflows.
Click the workflow you want to inspect.
In the details panel, review Run history, where each run shows its time and status, such as Completed.
Click a run to open its Run log, including the start time, end time, and runtime.
The Activity tab under Workflows in your app dashboard brings together runs from all your workflows in one place.
Runs you start with Run now are labeled TEST, so you can tell them apart from runs that fired automatically.
To view activity across workflows:
Click Dashboard in your app editor.
Click Workflows.
Click the Activity tab.
Review the list of runs, where each one shows its workflow, status, start time, and run time. Filter by status (Running, Completed, Failed, or Cancelled) or by time range.
Click View on a run to open it in the workflow’s diagram alongside its run log.
The Activity tab, showing runs across all your workflows.
Workflows are billed by the steps that run. Each step that runs a backend function uses a fraction of an integration credit, and the charges add up across all the steps in a single run. Built-in integrations that a step uses, such as sending an email or calling an AI model, can use credits too, as they do elsewhere in your app.A workflow with several steps therefore costs more per run than a single-step one, so keep an eye on how often your workflows run and what each step does. If you run out of integration credits while a run is in progress, that run is cancelled. Learn more about credits.
Tips for managing credit usage
Workflows that run frequently or chain several actions can drain your credit balance quickly. Here are some ways to keep usage under control:
Add specific trigger conditions: Narrow the criteria that start a workflow so it does not run unnecessarily.
Use conditions to skip unneeded steps: Branch the workflow so expensive steps only run when they are actually needed.
Consolidate redundant workflows: If several workflows perform overlapping tasks, combine them into fewer, smarter ones.
Resolve failing workflows promptly: Workflows that error can still consume credits on each retry. Check your run history regularly and fix or disable problem workflows as soon as you spot them.
Review the frequency of scheduled workflows: Consider whether recurring workflows, such as daily reports or reminders, need to run as often as configured.
Be sparing with AI calls: If a step calls an AI model, that uses credits too, so only call it when you actually need the result.
Test on a small dataset first: Run a workflow manually and check its run history to estimate the per-run credit cost before enabling it at scale.
Yes. Ask the AI chat to add conditions to any workflow or step. For example:
After the wait, only send the reminder when the task is still incomplete.
The workflow then takes different paths based on the conditions you describe, so different steps run depending on your data. A condition can have several paths, plus a catch-all for anything that does not match, and the diagram shows each path separately.
What happens when a workflow fails?
Open the workflow and check its Run history to find the step that failed. Ask the AI in your dashboard to explain the error and suggest a fix, then use Run now to confirm the workflow succeeds.If a workflow keeps failing, Base44 deactivates it automatically and emails you, so you can fix it before turning it back on. A scheduled workflow is deactivated after 5 consecutive failures, because it runs the same way every time, so repeated failures usually mean something is wrong. Workflows with any other trigger are deactivated after 50 consecutive failures, because their incoming data varies and the occasional failure is more expected. Fixing the workflow and reactivating it resets the count.
Why does Run now ask me to choose a previous run?
It depends on how the workflow starts. A scheduled workflow can run on demand at any time, because it does not need any input. A workflow that starts from an entity, connector, or in-app agent trigger needs the trigger’s data to run, such as the specific task that was created, so Run now replays the data from a previous run. This means the workflow has to run at least once before you can run it on demand.If you see No previous runs, trigger the workflow for real once, for example by creating a task for the task follow-up workflow, and then use Run now to replay that run.
Can I manage workflows from the AI chat?
Yes. Ask the AI to activate, deactivate, archive, or restore a workflow, and it makes the change for you without opening the dashboard menus.
