Customer story
Gépteszt Kft.’s inspection system, in production on ATAILA
A working application in two months; in production three months later. A web application, a field app and four private AI capabilities, on hardware in Hungary, with no external AI service.
- to a working application
- 2 months
- in production
- 3 months later
- private AI capabilities in daily use
- 4
- client organisations brought over
- 300+
- pieces of equipment brought over
- 20,000+
- inspection records brought over
- 150,000+
The customer and the system
Gépteszt Kft. has done occupational and machine-safety testing since 1987. It is an accredited lifting-equipment inspection body with an in-house ISO/IEC 17025 laboratory, and an EU notified body (NB 2233). With decades of records, it needed a modern application at the core of its business: the client organisations, their equipment and the inspections carried out on it.
What was built on the platform is the core inspection system: a web application for the office, a field app for the inspectors and private AI inside the workflow.
The legacy register was brought over: more than 300 client organisations, over 20,000 pieces of equipment and more than 150,000 inspection records. The office workflow was cut back to what the operation needs. The menu went from about 36 items to 11 for go-live, and every role was reviewed before launch.
How it went live
The team had a working application in two months, and took it into production three months later. The application from the second month is the one that went into production.
The field app was rolled out to the inspectors through Google Play. It was published from the customer’s own release process, and the app-store account and the signing keys are in the customer’s name.
Four private AI capabilities, inside the workflow
AI was part of the build from the first day, and since go-live it runs in production too. Every call is metered per feature, and the organisation can see the numbers.
- Nameplate identification
- Identifies a piece of equipment from a photo of its nameplate.
- Knowledge chat
- Answers questions over the company’s own documents.
- Visual similarity search
- Finds similar equipment from a photo.
- Category-icon generation
- Generates the icons for the app’s equipment categories.
The rule is fixed, and it holds for all four:
AI proposes, the portal decides, a human confirms.
The public face, rebuilt
Gépteszt’s website, gepteszt.hu, was rebuilt on the platform in two languages. It has a knowledge centre and carries structured data for search engines and AI answer engines. Visits are measured with first-party cookieless analytics, and the site ships with hardened security headers.
What the customer owns
Gépteszt owns its core system, in writing:
- The code, in its own source control.
- The data, in its own database.
- Its environments.
- The app-store account and the signing keys.
- The backups.
Published with the consent of Gépteszt Kft.
Planning a core system of your own?
Walk us through it, and we will show you how this one was built and what the first weeks would look like for yours.