Skip to content

New GenAI & AI agent development

Case study · AI & ML

Claims emails read and sorted by software

An insurer taught software to read incoming claim emails and route each one to the right handler.

  • Insurance
  • UK
  • 1,000+ employees
Industry
Insurance
Region
UK
Company size
1,000+ employees
Practice
AI & ML

The challenge

What was going wrong

The insurer received claim notices by email, with photos, forms and scanned letters attached. A small intake team read each message, worked out the type of claim and copied details into the core claims system by hand.

Messages were written in every style imaginable. Some mixed several claims in one email. Mistakes in sorting meant slow handoffs, and the team could not simply guess, because claims data is sensitive and regulated.

Our approach

How we worked it out

  1. 1

    Study the intake desk

    We watched the intake team work and collected anonymized sample messages. We listed every claim type and every field the claims system needs.

  2. 2

    Design with safeguards

    We chose a language model approach that flags low confidence cases for a person. Data stayed inside the insurer's own cloud boundary.

  3. 3

    Build and connect

    We built the reading and sorting service, then connected it to the claims system and the shared inbox through the existing APIs.

  4. 4

    Launch and tune

    We ran the tool beside the team first. Handler corrections were fed back to improve accuracy before it took over routine sorting.

What we built

The solution

  • Claim reader

    Extracts claim type, policy details and key facts from emails and attachments.

  • Smart routing

    Sends each claim to the right queue based on type, urgency and complexity.

  • Review screen

    Shows handlers the extracted fields and flags anything the model is unsure about.

  • Feedback loop

    Handler corrections are stored and used to retrain the models over time.

What changed

The results

  • Manual sorting reduced

    The intake team stopped reading every message just to decide where it should go.

  • Cleaner claim records

    Fields arrive in the claims system pre-filled, so fewer records need later correction.

  • People focus on judgment

    Staff spend their time on unusual claims and on customers who need a call.

Technology

What it was built with

  • Python Readable language for business logic, data and automation.
  • spaCy Production-grade library for text processing and entities.
  • Hugging Face Hub of open models and libraries for language and vision work.
  • FastAPI Python framework for fast, well-documented APIs.
  • Azure Microsoft cloud hosting and services.
  • PostgreSQL Dependable database for complex business data.

Keep reading

More case studies

All case studies
Manufacturing Europe

Defects caught at the station, not at the customer

Cameras catch surface defects before shipping

Problem
Inspectors checked parts by eye and missed small flaws.
What we did
Trained vision models on real parts and mounted cameras on the line.
What changed
Flawed parts are flagged the moment they pass the camera.
  • Python
  • PyTorch
  • OpenCV
  • ONNX
Read the story
SaaS & Technology USA

Customers get sourced answers without waiting

A support assistant that answers from real docs

Problem
Support answered the same questions again and again.
What we did
Built a retrieval based assistant grounded in product docs.
What changed
Customers get answers with links to the source article.
  • Python
  • LangChain
  • Qdrant
  • FastAPI
Read the story
Energy & Utilities Asia-Pacific

Crews hear about faults before customers do

Remote sensors warn crews before equipment fails

Problem
Remote equipment was checked by driving out on a fixed schedule.
What we did
Fitted sensors and built a live monitoring dashboard.
What changed
Crews are sent where readings show a real problem.
  • MQTT
  • Raspberry Pi
  • Espressif ESP32
  • TimescaleDB
Read the story
Taking new projects

Let’s talk about your Insurance project

Tell us about your project and get a free consultation and estimate within 24 hours.

  1. We reply within 4 business hours A real consultant reads your brief — no bot, no queue.
  2. 30-minute discovery call We dig into goals, constraints and budget. NDA signed on request.
  3. Proposal in 24 hours Scope, timeline, team and price — clear enough to decide on.

NDA on request You own the IP No obligation Free estimate

Your information is confidential. We’re happy to sign an NDA.

Book a free call Chat on WhatsApp