AppSheet
Web PlatformRead and write the data behind your Google AppSheet apps. Agents add, edit, delete, and find rows in any table or slice, honoring the app's security filters and firing the bots you have configured for each action.
What This Integration Enables
The thing that separates AppSheet's API from a raw database connection is that it goes through the app: every operation fires the workflows and bots you configured, honors the app's security filters, and can impersonate a named user for audit attribution. That is precisely what makes it safe to automate. FlowRunner agents drive the table Action API, adding, editing, deleting, and finding rows in any table or slice, and the app behaves exactly as if a person had made the change inside it. - Push orders, form submissions, and CRM records into AppSheet tables as new rows - Keep tables in sync with external systems on schedules or events - Read filtered, sorted, limited data out with AppSheet expression Selectors - Fire the app's configured bots on every automated change, not around them - Impersonate specific users so security filters and audit attribution apply Deletes are the operation that removes working data from a team's hands, and that is where FlowRunner's [human-in-the-loop](/concepts/human-in-the-loop) gate stands.
Without FlowRunner
With FlowRunner
Use Case Scenarios
Orders reach the field app before the field asks
A [Shopify](/integrations/shopify) order fires the workflow, and Add Rows writes it into the operations table with key columns auto-generated by AppSheet. The add bot the team already built sends its usual assignment notification, because the API write triggers it like any in-app add. Field staff open the app to find the order already there, already assigned.
Payment confirmation closes the loop
When [Stripe](/integrations/stripe) confirms a payment, the agent runs Find Rows with Filter("Orders", [Status] = "Open") scoped to the matching reference, then Edit Rows with the key column and the paid status. The edit bot updates the customer-facing view. What used to be a twice-daily reconciliation pass becomes an event that handles itself, with the mismatches, payments matching no open order, posted to [Slack](/integrations/slack) for a person.
The reporting export that honors the app's permissions
Nightly, the agent pulls open records with a Top(Filter(...), 500) Selector, running as a reporting user whose security filters limit what the export can see, and appends the rows to [Google Sheets](/integrations/google-sheets). The report inherits the app's access model instead of bypassing it, which means the export shows exactly what that role is allowed to show, by construction.
Human-in-Loop Highlight
Delete Rows needs only a key column value to remove a record, and because the API drives the app's real Action surface, each deletion also fires whatever delete bots the app owner configured: notifications go out, downstream updates cascade, and the row disappears from the app field staff are holding in their hands. A cleanup agent that guesses wrong does not fail quietly. So deletion batches are assembled, never fired: the agent gathers candidates with Find Rows, includes each record's contents and age, and posts the proposal to the app's owner, "43 completed work orders older than one year, delete bots will notify the archive workflow, proceed?" The supervisor who approves knows what the bots will do because they built them. Adds and edits run unattended, because the app's own rules validate them; removal of records a team may be mid-task on is a call a person makes.
Agent Capabilities
4 actionsRecords
4- Add Rows Adds one or more rows to a table via the Add action, mapping column names to values with auto-generated keys omitted. Runs any configured add bots and returns the created rows with computed and key values. Supports Run As User Email for security filters and audit attribution.
- Edit Rows Updates existing rows via the Edit action. Each row must include the table's key columns plus the columns to change. Runs configured edit bots and returns the updated rows, with optional user impersonation.
- Delete Rows Deletes rows identified by their key columns via the Delete action, firing any configured delete bots. The operation this page's human gate exists for, with impersonation support so removals attribute correctly.
- Find Rows Reads rows via the Find action, returning everything the app's security filters allow or a set shaped by an AppSheet expression Selector: Filter, OrderBy, Top. Can also fetch specific key rows. The read behind queues, reconciliation, and reporting.
Frequently Asked Questions
What can FlowRunner do with AppSheet?
FlowRunner agents can run Add Rows, Edit Rows, and Delete Rows in AppSheet, plus 1 more action.
Does connecting AppSheet to FlowRunner require OAuth?
No. AppSheet connects to FlowRunner with an API key, no OAuth flow required.
Can AppSheet trigger a FlowRunner workflow automatically?
AppSheet doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Start building with AppSheet
$100 in credits. No card required. Connect in minutes.