CASE-03
Finance & Insurance
Generative AI & Agents
Sample scenario ยท figures to be confirmed
AI document analysis: speeding up claims processing
- CLIENT
- [Client name] โ insurance brokerage firm, [150] employees
- INDUSTRY
- Finance & Insurance
- DURATION
- [4 months]
- EXPERTISE
- Generative AI & Agents
- FORMAT
- Prototype
5 min
pre-analysis time (vs [45 min])
+40%
more cases processed per handler
0
data transferred outside the EU
The challenge
Every claim file contains around fifteen different types of documents: reports, invoices, photos, letters. Handlers spent most of their time reading, sorting, and re-keying data.
OUR APPROACH
V-U-L-Y-X approach
V
VisionPrioritizing the 3 most frequent claim types.
U
Unified IntelligenceAutomatic retrieval of documents from email and the case management tool.
L
LogicExtraction with systematic source citation; mandatory human validation (AI Act).
Y
YieldMeasuring processing time before/after on [500] case files.
X
eXperimentationPilot with a team of 4 volunteer handlers.
The solution
When a case is opened, the handler gets a pre-filled summary sheet, a list of missing documents and points of attention, with every piece of information linked back to its source document.
TECHNOLOGIES ยท Document extraction (PDF, photos, emails), LLM hosted in Europe, integration with the case management tool
SCREENSHOT ยท ARCHITECTURE DIAGRAM
What's next?
Extension to underwriting files.
Start a similar project โ