← All projects
Selected project / Insurance Sherlock
Make complex claim evidence easier to inspect.
A prototype combining document forensics, image checks and AI-assisted interpretation to organise evidence for insurance claim review.
01 / The challenge
Insurance Sherlock
Invoices, reports and photographs can contain inconsistencies that are hard to see through manual reading alone. Investigators need findings they can check against the source.
The engineering approach
Local PDF and DOCX checks inspect metadata, timestamps, structure and embedded content. Image analysis and LLM workflows add context, while rules organise indicators into a reviewable report. Indicators guide investigation; they do not establish fraud.
Source
PDF / DOCX / JPEGEvidence review
01PDF & DOCX inspection
02Metadata & image checks
03LLM evidence synthesis
04Structured review reports
02 / System architecture
Make complex claim evidence easier to inspect.
Local PDF and DOCX checks inspect metadata, timestamps, structure and embedded content. Image analysis and LLM workflows add context, while rules organise indicators into a reviewable report. Indicators guide investigation; they do not establish fraud.
Source
PDF & DOCX inspectionMetadata & image checksLLM evidence synthesisStructured review reports
ReviewHUMAN
DECISION
DECISION
03 / From input to useful output
What the system connects
Collect evidence
PDF & DOCX inspection
Inspect files
Metadata & image checks
Combine findings
LLM evidence synthesis
Investigator review
Structured review reports
Contact
Working on a similar challenge?
Tell us about the workflow, the constraints and the result you need.