How to Automate Imports and Deliveries with Workflows

What a Workflow Is

A workflow is one pipeline with three parts:

1. Input: where documents arrive. A folder in a connected drive, or email forwarded to your import address.
2. Parsing: what the documents are. Bank statements, or invoices and receipts.
3. Output: where the result goes. A file written to a drive, a webhook to your own system, or both.

Both ends can be lists. Statements might arrive in two shared folders and by email, and the result might go to a Google Sheet and a webhook. You can have as many workflows as you need, such as one per client.

Everything a workflow imports also lands in Bankflow as usual, on the Statements page or in Convert, and any API webhooks you've set up still fire. A workflow's outputs are extra copies.

Step 1: Create a Workflow

Open Workflow in the sidebar and click New workflow. Give it a short name, like "Statements from the shared drive". The editor lays the pipeline out left to right.

Step 2: Add Inputs

A folder

Click Add a folder, choose a connected account, and browse to the folder to watch. Leave it as The whole drive to watch everything. If you haven't connected an account yet, the editor asks you to; see connected accounts. Folders can be read from Dropbox. Google Drive can only be written to, so it can't be an input.

Watched folders are checked every 15 minutes. The first check imports what's already in the folder, then each later check picks up only what's new. Every file imported uses credits as normal, so point a new workflow at a folder that holds only what you want read.

Forwarded email

Click Forwarded email. The card shows your import address, and lets you narrow which mail this workflow takes: a particular approved sender (or Any approved sender), and optionally text the subject must contain. Matching mail is parsed as soon as it arrives instead of waiting in your email inbox. See importing by email.

• Only mail from approved senders ever reaches a workflow. The sender filter decides what happens to mail, never who may send it.
• Each email goes to one workflow. If two could take it, the more specific one does.

Step 3: Choose How Files Are Read

• Bank statements: transactions and balances, into your ledger. Picks up PDFs and spreadsheets (CSV, XLSX, XLS).
• Invoices and receipts: line items and totals, into Convert and kept out of your ledger. Picks up PDFs and photos (JPG, PNG, WEBP), including photos emailed from your phone.

Files of other kinds in a watched folder are skipped.

Step 4: Add Outputs

Write a file

Click Write a file and choose where it goes:

• To: a connected account, or Where it came from to write back to the drive the file was read from. Forwarded email has no folder to write back to, so for an email input, choose an account.
• Folder: beside the original, or a folder you pick. The file is written into a Parsed subfolder there. That subfolder is kept out of the watch, so the workflow never re-imports its own output. For Google Drive, type a folder name; Bankflow creates it.
• Format: CSV, Excel, JSON, or, when the destination is Google Drive, a real Google Sheet you can open and edit.
• Columns: the same fields and presets as a manual export. For invoices and receipts, also choose one row per document or one row per line item.

If a file is parsed again, the new result is added beside the old one, never written over it.

Post to a webhook

Click Post to a webhook and enter an https address on the public internet. Each result arrives as a workflow.completed POST with the parsed rows as JSON, signed with an X-BankFlow-Signature header exactly like the API webhooks. Click Generate secret to get the signing secret now, so you can set up your receiver, or leave it and one is created when you save.

Remove from Bankflow once sent

If the workflow hands statements to another system and you don't want a copy kept here, turn on Remove from Bankflow once sent. The statement and its transactions are removed only when every output was delivered; if any fails, nothing is deleted. Credits are still charged and the billing record is kept. It needs at least one output, or nothing would be left of the import.

Running and Monitoring

The Workflow page lists each workflow with where it reads from, where it sends to, and when it last ran. Across the top are the number of workflows running, files imported, files written back, and anything that needs attention.

• Run now checks a workflow's folders immediately instead of waiting for the next check.
• Switch off pauses a workflow without deleting it; Switch on resumes it.
• Open shows the workflow's setup and its history: every file it handled, whether it was imported, and whether it was written to each destination.

Select several workflows to run, switch on, switch off, or delete them together. Deleting a workflow keeps everything it already imported; only the pipeline goes.

When Something Goes Wrong

A file was imported but not written back. Its history row says Not written, with the reason. Click Retry. Only the destinations that refused it are tried again, and the file isn't parsed or charged again.

"Its accounts need reconnecting before this can run." A connected account's access expired or was revoked. Reconnect it from connected accounts, and the workflow picks up where it left off. If only some of a workflow's folders can't be read, the others keep importing.

"Stopped: out of credits." Top up, and the rest is picked up on the next check. Nothing is skipped.

"This has nowhere to read from." Edit the workflow and choose a folder or add an email input.

Next Step

Connect the drives your workflows read from and write to in connected accounts.