InsuranceHIPAA · SOC 2 · GDPR

From claims chaos to structured certainty.

ACORD forms, loss runs, medical attachments, policy applications, and FNOL packets. Bem classifies, splits, extracts, and routes the data into Guidewire, Duck Creek, or your own claims platform. HIPAA-compliant pipeline. End-to-end encryption. Zero retention on request.

Shipping with PromptWell and Carrot Fertility. First 100 function calls every month, free.

Documents we handle

Every form your back office actually receives.

Insurance documents are the worst-case version of every document problem. Multi-page packets, mixed types, handwritten adjuster notes, scans, faxes, carrier-specific layouts. Bem ingests all of it.

01 / 06
ACORD forms

ACORD 25, 125, 130, and the long tail. Field-level extraction across every carrier variant.

02 / 06
Loss runs

Multi-page carrier loss histories. Per-claim breakouts, frequency, severity, open vs closed.

03 / 06
Medical records

Provider notes, EOBs, diagnostic reports, prescription histories. HIPAA-compliant pipeline.

04 / 06
Policy applications

New business submissions. Risk classification, declared exposures, prior coverage.

05 / 06
FNOL packets

First notice of loss, claim forms, police reports, adjuster notes. Split, classify, route.

06 / 06
Subro and recovery

Demand letters, settlement docs, supporting evidence. Structured for recovery workflows.

The workflow

A 47-page packet. One workflow.

The hardest part of insurance automation is not the extraction. It is the multi-document packet that arrives as one file. Bem decomposes the packet, classifies each record, and routes the structured output to your claims system with the full audit trail intact.

Packet decomposition
SPOTLIGHT

Split, classify, extract. In one call.

Bem takes a 47-page claims packet and returns a structured record per document, with every confidence score, every audit anchor, every routing decision logged. Your team reviews exceptions, not every field.

Split
Semantic packet decomposition
Classify
Per-document carrier and type
Extract
Type-safe fields with scores
A worked example

A claims packet, structured.

POST /v3/functions/extract.run200 OK · 4.1s
{
  "claim_id": "C-2026-0428-991",
  "claimant": {
    "name": "Daniel R. Carmichael",
    "policy": "PRX-487-A"
  },
  "incident": {
    "date": "2026-04-21",
    "type": "auto_collision"
  },
  "documents": 47,
  "confidence": 0.972
}
Type-safe, confidence-scored, audit-anchored. Ready for Guidewire.
01
Classify

Identify document type across an entire multi-page packet. ACORD vs loss run vs medical, by carrier and line of business.

02
Split

Decompose a 47-page claims packet into its constituent records. Each one labeled, each one routed to the right extractor.

03
Extract

Policyholder, claim details, coverage limits, line items. Layout-aware models, confidence-scored fields.

04
Enrich

Resolve against policy systems, prior claim history, provider directories, and your fraud indicators.

05
Payload Shaping

Format the final structured output for Guidewire, Duck Creek, Majesco, or whatever your claims platform expects.

Customer · PromptWell

“Before bem we had three engineers full-time on the data-format problem. Now the schema is the contract, and the carrier-specific variation lives inside bem, not in our codebase.”

80%
reduction in engineering time handling data formats at PromptWell. Carrot Fertility uses the same pipeline for claims processing.
HIPAA · SOC 2
Build in minutes

Build in minutes. Start for free.

Whether it's our no-code builder or APIs, ship your first workflow in minutes, not days. Get the first 100 function calls for free, then pay-as-you-go with graduated pricing.

bem for Insurance | Claims, ACORD forms, loss runs | bem