Technical Overview

How Adjustsage processes a claim

A step-by-step walkthrough of what happens between document receipt and handler assignment. Four stages, under 4 seconds.

Step 01 / Ingestion

Any source, any format

Adjustsage connects to your existing intake channels via API, IMAP listener, or webhook. Every inbound channel feeds a single normalized event queue. Documents are fingerprinted, deduplicated, and queued for extraction.

Supported formats: PDF (all versions), TIFF, JPEG, PNG, plain text, HTML email body, EDI 837, JSON portal payloads. Attachments within attachments are recursively extracted.

email_imap

enabled: true

portal_webhook

enabled: true

edi_837

enabled: true

fax_ocr

enabled: true

dedup_window

24h rolling

policy_number

HO-2847391

insured_name

J. Whitmore

loss_date

2026-06-14

loss_type

water_damage

loss_location

44 Birch Ave, CT

third_party

none declared

Step 02 / Extraction

32 fields resolved in under a second

Named-entity recognition trained on 4+ million real P&C claims identifies every structured field in the document. Each field receives a confidence score. Fields below the carrier-defined threshold are flagged for human review.

Extracted fields include: policy number, insured name, loss date, loss type, loss description, loss location, reported parties, witness data, attorney representation flag, prior claims indicator, and 20+ additional fields.

Step 03 / Coverage Mapping

Coverage matched before the handler ever touches the file

Adjustsage queries your policy system API with the extracted policy identifier. It maps the loss type against the applicable coverage section, pulls deductible and limit values, and identifies any exclusions that may apply.

The coverage match result is included in the structured claim event so your handler sees it the moment they open the file, not after 10 minutes of manual lookup.

coverage_section

98%

policy_limit

100%

deductible

96%

exclusions_check

91%

effective_date

100%

queue

fast_track

reserve_band

$8k-$12k

fraud_score

0.18 (low)

complexity_flag

standard

reason_code

R-14: domestic water

Step 04 / Routing

Routed to the right handler, every time

The completed structured claim event is scored for complexity and fraud indicators, then routed to the appropriate handler queue via your claims system API. Every routing decision includes a machine-readable reason code and a full field audit trail.

Routing rules are configured per carrier and updated without code changes. Complex or flagged claims can be escalated to a supervisor review queue before touching a handler.

Technical FAQ

Common integration questions

Does Adjustsage replace my claims management system?

No. Adjustsage is a pre-processing layer that sits between your intake channels and your existing CMS. It handles ingestion, extraction, coverage lookup, and routing, then pushes a structured event to your CMS via API. Your handlers work inside the system they already know.

How long does integration take?

Standard integration with a supported CMS (Guidewire, Duck Creek, Applied Epic) takes 2 to 4 weeks. Custom or proprietary systems take 4 to 8 weeks depending on API availability and policy data schema complexity. We provide an integration engineer for the duration.

What happens when the AI is not confident about a field?

Fields below the confidence threshold (configurable, default 85%) are flagged in the structured event with a low-confidence marker. The claim is still routed, but the handler sees the specific fields that need manual verification highlighted in their view. We do not silently fill uncertain fields with best-guess values.

How does Adjustsage handle claims with no policy match?

Claims where the policy lookup returns no match are routed to a carrier-defined exception queue with a reason code indicating the lookup failure. The structured event still includes all successfully extracted fields so the handler can complete coverage lookup manually with most of the data already populated.

Is data retained after processing?

Processed claims data is retained for 90 days by default to support audit and dispute resolution. Retention period is configurable under the Production tier. Raw document files are not retained after extraction unless carriers opt into an archival add-on. All data is encrypted at rest (AES-256) and in transit (TLS 1.3).

See the pipeline on your own documents

We run a free sandbox session with a sample of your FNOL documents. No integration, no commitment.

Request a Sandbox Full Platform Overview