WorkerKitJoin the waitlistAI workers for Extraction
AI workers for extraction pull specific fields out of documents, emails, and attachments into structured data. WorkerKit controls which files they may open and read.
- Email OTP Retrieval Language model Find the latest one-time passcode email for an agent sign-in without changing or sending mail.
- School Email to Calendar Language model Scan school emails and create complete calendar events organized by child name.
What is a Extraction AI worker?
A extraction AI worker is an AI agent set up to run the extraction job unattended, reading a document, email, or attachment and pulling named fields out into structured data. It typically touches Gmail and Outlook attachments, Google Drive and SharePoint files, and CRM records, reading only the files and fields you have approved and writing the extracted values where you choose.
What these workers do
Extract totals from Outlook invoices
Reads PDF invoices attached to Outlook mail and pulls vendor, date, and amount into a structured row for the finance team's spreadsheet.
Pull fields from Drive contracts
Opens contract files in Google Drive and extracts party names, term dates, and renewal clauses into a clean summary record.
Capture details from signup emails
Reads inbound Gmail signups and extracts name, company, and request into structured fields ready to drop into a HubSpot record.
Read line items off receipts
Extracts merchant, date, and line-item totals from receipt attachments so an expense record is built without manual typing.
How a Extraction AI worker runs
What it connects to
When it runs
What stays under your control
Extraction questions, answered.
How is an extraction AI worker different from enrichment?
Extraction pulls fields out of a document you already have, enrichment looks up new facts from outside to add to a record. An extraction worker reads an invoice and captures its total. An enrichment worker instead finds a company's industry to fill a blank CRM field. One reads what is there, the other fetches what is missing.
Can an extraction worker read the full contents of a file?
It reads the file body you grant it, and you can redact parts it should not see. An extraction worker needs to open the document to pull fields, so WorkerKit lets you scope it to specific folders and redact sensitive sections first. It reads only within that grant and writes just the fields you asked for.
Where does an extraction worker put the data it pulls?
It writes the extracted values to the destination you choose, a spreadsheet, a record, or a structured output. An extraction worker turns an unstructured invoice or contract into clean fields and drops them where you point it. WorkerKit keeps it read-only on the source, so extracting never alters the original file.
Will an extraction worker handle scanned or PDF attachments?
It reads the attachments you grant it from Gmail or Outlook, pulling named fields from their contents. An extraction worker works on the documents you scope it to and captures the values you specify. You control exactly which mailboxes and files it may open, and every file it reads is logged for review.