Skip to content
← All projects
Automation2026

Data Collection Automation for an Insurance Partner

Repair company partnered with insurers

An RPA platform that collects accident and vehicle data and photos that staff used to copy by hand from insurer portals each day, and feeds them into the internal system.

Deliverables
RPA workersCentral management serverAdmin screens
Tech
PythonPlaywrightFastAPIMySQL

Problem

Every day, staff logged into insurer portals, looked up claims by accident number, copied vehicle details into the internal system, and downloaded photos into set folders. The more cases there were, the longer it took, and copying by hand led to errors.

Constraints

  • Portal login required a digital certificate and a secure keyboard, so we first had to prove automation was possible at all.
  • Bad data written straight into the production database is hard to undo.
  • Each insurer's portal looks different and changes often.
  • It would start on one PC but was planned to grow to 6–10.

Solution

  1. 1

    One week before main development went to a proof of concept: we measured the authentication method and session lifetime and confirmed that login and lookup could be automated. This avoided spending 8 weeks without knowing whether it would work.

  2. 2

    Collected data goes into a staging table, not straight into production tables. Only records an admin reviews and approves move into production data.

  3. 3

    Each insurer's module is separate from the common engine, so when one portal changes, only that module needs fixing.

  4. 4

    A central management server distributes jobs, monitors each PC's status and run history, and retries failed jobs.

  5. 5

    Photos are saved to the company NAS in automatically created year/month/insurer/date folders.