FlowRunner
PricingContact
Theme
Start Free

Doc Crafter

Documents

Doc Crafter is a stateless document toolbox. Agents turn web pages and spreadsheets into PDFs, read text out of PDFs and photos, generate and decode QR codes and barcodes, and merge or split PDF files in single calls.

13 actions API key available
Platform Documentation Capability data verified 2026-08-13
The monthly statement run arrives as one printed PDF in file storage
Split PDF By Text separates it wherever the statement header phrase appears
Agent compares the number of documents produced against the number of accounts billed
Extract Text From PDF reads the account name off the first page of each piece
Each name is matched against the billing list and every mismatch is collected
The count, the mismatches, and one sample document post to the billing owner
A person confirms the split landed on the right boundaries before any statement reaches a customer

What This Integration Enables

Doc Crafter is a toolbox, not a platform. There is no account state, no template library, no identifier for anything, and nothing is stored on the vendor's side. Every operation is one stateless call: give it an address, get a result back, which FlowRunner puts into file storage. That shape makes it the wrong tool for producing branded documents from a designed template, and the right tool for the unglamorous middle of a process, where a batch has to be separated, a scan has to be read, or a code has to be produced or decoded.

FlowRunner agents use it in both directions. Outward, it turns a web page or a CSV into a PDF and produces QR codes and barcodes for tickets, labels, and work orders. Inward, it reads text out of PDFs and photographs, and reads codes back off images taken in a warehouse. Because it fetches whatever address it is given, a flow should point it at file-storage URLs from earlier steps rather than at anything confidential, and that is a discipline worth stating out loud rather than assuming.

Without FlowRunner

Separating a batch by hand Somebody opens the printed run and saves out one document per customer
Scanned paper as a dead end A delivery note arrives as an image and has to be read and retyped before anything can act on it
Labels made in another tool A QR code for a work order is generated somewhere else and pasted in

With FlowRunner

Separating a batch by rule The run splits on the phrase that starts each statement, and the count is checked against the ledger
Scanned paper as data Character recognition turns the delivery note into text a flow can match against the purchase order
Labels made in the flow The code that points at the record is generated in the same run that produces the document

Use Case Scenarios

  • Reading what arrived on paper

    A supplier delivery note arrives as a photograph from a warehouse phone, landing in Amazon S3 or Google Drive. The agent calls Extract Text From Image, matches the extracted line items against the open purchase order, and posts the differences. Where a scanned PDF arrives instead, the agent tries Extract Text From PDF first and falls back to Extract Text From Scanned PDF when the plain extraction comes back empty, because character recognition is slower and there is no point paying for it on a document that already contains text. The receiving clerk reads a comparison rather than a document.

  • The work order that carries its own identity

    When a job is created in Airtable or the operational system of record, the agent calls Generate QR Code with a value that points back at that record, then CSV To PDF or HTML To PDF to produce the printed work order with the code on it. Later, a technician photographs the code on the shop floor and Read QR Code turns the photograph back into the record identifier, so the completion event attaches to the right job without anybody typing a number. Generate Barcode covers the same loop where the label has to match an existing symbology, and Read Barcode closes it.

  • One run, many customers

    The billing system produces the month's statements as a single PDF. The agent calls Split PDF By Text on the phrase that opens each statement, which turns the printed run into one document per customer, and stores the pieces. Merge PDF Files handles the opposite case, joining a cover letter, an invoice, and an appendix into one file before it goes out through Gmail. Split PDF By Page covers the fixed-layout case where the boundary is a page number rather than a phrase.

Human-in-Loop Highlight

Split PDF By Text has a failure mode that looks exactly like success. It splits wherever a phrase appears, so if the statement header wording changed this month, or a customer's own text happens to contain the phrase, the split lands on the wrong boundaries. The call still returns cleanly. The archive still contains documents. Every document still opens, still paginates, and still looks like a statement. The only thing wrong with it is that the last page of one customer's statement is now the first page of another customer's, and if the flow sends them, that is a disclosure of one customer's balance to another. Nothing in the response can tell you this happened, because from the vendor's side nothing did. So FlowRunner agents never let a split feed a send unattended. The agent counts the pieces, extracts the account name off the first page of each one, checks those names against the billing list, and posts the result: "The March run split into 418 documents. 421 accounts were billed. Three documents contain two different account names on the first page. Sample attached. Send, or reprocess?" A person looks at the sample and decides. This is the digital andon cord in its plainest form: the flow stops itself on a signal that everything downstream would have read as fine.

Agent processes routinely
Detects exception requiring judgment
Clear match Continues automatically
Ambiguous Routes to human via preferred channel
Human decides
Agent resumes with decision

Agent Capabilities

13 actions

Generate

4
  • HTML To PDF Turns a web page into a PDF and stores it in file storage. Margins are given in millimeters, and zero really does mean a borderless page.
  • CSV To PDF Turns a CSV file into a formatted PDF table and stores it in file storage, which is how a query result becomes something a person can be handed.
  • Generate QR Code Turns text into a QR code image and stores it in file storage. Used to put a pointer back to the record on a ticket, a label, or a work order.
  • Generate Barcode Turns text into a barcode image and stores it in file storage. Eleven symbologies are available and the data has to suit the one chosen.

Extract

5
  • Extract Text From PDF Reads the text out of a PDF that already contains text. The cheaper and faster of the two PDF readers, and the one to try first.
  • Extract Text From Scanned PDF Reads the text out of a scanned PDF using character recognition. Slower than the plain extraction, and worth naming the language.
  • Extract Text From Image Reads the text out of a photograph or scan, which is how paper arriving from the floor becomes something a flow can act on.
  • Read QR Code Reads a QR code out of an image and returns what it contains, closing the loop from a printed document back to its record.
  • Read Barcode Reads a barcode out of an image and returns what it contains.

Files

3
  • Merge PDF Files Joins several documents into one PDF and stores it in file storage. Takes a list of addresses in page order, or a single zip containing them, never both.
  • Split PDF By Page Pulls out the pages named and stores the result in file storage. The right choice when the boundary is a fixed layout rather than a phrase.
  • Split PDF By Text Splits a PDF wherever a phrase appears, which is how a printed run of statements becomes one document per customer. This is the operation this page's human gate exists for, because a wrong boundary produces a document that still looks correct.

Account

1
  • Check Credentials Confirms the key works, which is how a connection is verified without converting anything.

Frequently Asked Questions

What can FlowRunner do with Doc Crafter?

FlowRunner agents can run HTML To PDF, CSV To PDF, and Generate QR Code in Doc Crafter, plus 10 more actions.

Does connecting Doc Crafter to FlowRunner require OAuth?

No. Doc Crafter connects to FlowRunner with an API key, no OAuth flow required.

Can Doc Crafter trigger a FlowRunner workflow automatically?

Doc Crafter doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with Doc Crafter

Free plan, no card required. Connect in minutes.