Case study
Handling 50% more cases without rewriting the calculation engine
For Fiabilis, technology is the production line. Having reached 6,000 cases a year, the infrastructure hit its limit. The answer: a cloud architecture that orchestrates the workflows and keeps the original calculation engine intact.
3 min read

Proof
−80% monitoring time per job
+50% client companies served, without growing the team
+50% profitability per client
one month earlier for reports to public bodies and collections
The context
The IT infrastructure of service companies runs an industrial production line: it extracts data, calculates, verifies, files. At the heart of the system there is often proprietary software that does the maths with surgical precision but cannot orchestrate workflows. If the surrounding architecture cannot absorb growth, operational capacity sets the ceiling on revenue.
When the production line is the revenue ceiling
Fiabilis, the Italian subsidiary of a group operating in eight European countries, recovers grants and incentives for businesses under a model tied to the amounts recovered. Processing speed determines billing. Every year 6,000 jobs pass through institutional portals, algorithms and human checks, and no single step restarts on its own.
The system requires 16 hours a day of manual monitoring and takes up a third of the IT team just to watch over the processes. The challenge: multiplying volumes without touching the calculation engine, where the intellectual property lies.
Move the control, don't rewrite the calculation
Fiabilis chose to separate workflow orchestration from the calculations themselves. To do so it brought in mediaengine, drawing on its track record of cloud architectures built around complex legacy systems.
Before defining the solution, the business case was broken down into its financial drivers. The analysis revealed an essential truth: the ability to scale did not depend on the language of the core software, but on how the queues were managed. The architectural choice was straightforward: take control of the process out of the engine and let the engine do what it was designed for, which is calculating. The trade-off is explicit: the legacy code stays untouched and the budget goes where the efficiency lies.
The flow moves forward on its own
and only stops for people
Isolating workflows from the calculation code.
An event-driven microservice architecture on Azure and Kubernetes makes it possible to add new steps to the production line without touching the core engine.
Automating workloads while keeping people in control.
Jobs progress overnight and restart automatically after an interruption, stopping only for the consultant's validation. The constraint on access to public administration portals was turned into an architectural rule: one dedicated queue per credential to guarantee parallelism.
Scaling in full compliance.
The infrastructure scales capacity according to queue depth, while centralised tracking meets GDPR and NIS2 requirements.
Ensuring a consistent interface.
We complemented the platform with a token-based Design System and delivered the infrastructure fully described as code in the repositories.
The ceiling rises, the pace stays the same
The infrastructure and the Design System are already in Fiabilis' hands: building on them, four developers with no front-end experience released the first version and now run the applications on their own. From here, growth is measured against the agreed drivers: queue orchestration aims to cut monitoring time per case by 80%, brings reports to public bodies forward by a month and makes previously excluded jobs profitable. With the same pressure on the team, the work handled in parallel multiplies. The next step, already planned, will introduce automatic data extraction from payslips.