Extract data from ACORD forms

The applications and section forms (125, 126, 127, 130, 131, 140, 160, and the 80 and 90 personal lines applications), the certificates and evidence forms (25, 27, 28), the service forms (35, 37, 45), plus loss runs, statements of values, supplementals and schedules. Any edition: fields are read by what they are, not where they sit, so the record comes out the same whichever edition the agency used.

Yes. LLMWhisperer keeps the form's layout, so each value is read from the box it belongs to, and anything uncertain goes to a person with the page rather than a guess going into the AMS.

Parse first, then extract. LLMWhisperer turns the PDF into layout-preserved text so every value stays tied to its box, and Document Studio maps that text to your schema. For bulk ACORD form processing, run it as an ETL pipeline, or hand Flow the agency inbox so submissions are picked up as they land.

Yes. Document Studio deploys an endpoint per document type that will extract ACORD form data on demand — POST the file, get the record back as JSON. The same extraction also runs as an ETL pipeline if you would rather drop files in a folder than call an endpoint.

Most ACORD form automation tools fill forms in; these read them. Unstract is three: LLMWhisperer reads the form and keeps its layout, Document Studio uses LLMs to turn that into the fields you named, and Flow runs the intake around them. The OCR layer is not an LLM — it is a parser built for LLM pipelines, which is why the output is reliable enough for the extraction step to trust.

Yes. Describe the intake in plain English and Flow's agents build the workflow — capture from the agency inbox, parse, extract, check across the forms, and post to your AMS. You approve the workflow before it runs, and it then runs the same way every time, stopping for an underwriter only where you said so.

Yes: premises on the 125 against the 140, prior carrier and loss history against the loss runs, NAICS against the operations described, and whatever else your underwriters check by eye today. A check that fails stops the submission with the reason attached.

The submission stops in a review queue, where an underwriter or account manager sees the field and the box it came from, then approves, corrects or sends it back. The decision is recorded against the submission and the rest of the inbox keeps moving.

Both. ACORD forms processing happens in layers here: LLMWhisperer is the ACORD forms parser and Document Studio returns the structured fields, while Flow is the submission intake automation around them — collecting submissions, running checks, routing exceptions and posting the result to your AMS, rater or policy system.

Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant: documents are encrypted in transit and at rest, never used to train models, and retained only as long as you choose. Self-hosted options are available for carriers that need processing inside their own environment; details are in the Trust Center.

Not for Flow: in a consultation we build the intake around your agencies' submissions, your checks and your AMS, and most teams run their first live intake within days. To try extraction today, start for free and upload a 125.