Featured Case Studies

Private healthcare providers must publish medical reports to the Electronic Health Record of several Italian regions, each with its own rules and infrastructure. We built a single gateway for FSE 2.0, integrated with 8 regions: Lombardy, Piedmont, Liguria, Umbria, Tuscany, Apulia, Lazio and Veneto.

The gateway handles validation against the national gateway, publication, replacement and cancellation of documents, metadata, confidentiality settings and time stamping, with a dedicated adapter for each region.

Private providers’ schedules and bookings must be exposed on regional booking systems. Our integrations with the booking systems of more than six regions publish schedules and services, receive and manage bookings, report outcomes and delivered services, and handle payments and receipts.

A web console lets operators monitor every flow and act on anomalies.

For companies working on roof renovation, asbestos removal and photovoltaic systems, finding the industrial buildings to work on still means site visits and manual research. We developed a system that analyses entire areas from satellite imagery and returns a list of roofs that are already measured, classified and linked to the businesses occupying each building.

A computer vision pipeline detects building outlines by matching the map with the satellite image, splits them into roof sections, calculates area and perimeter in metres from geographic coordinates and crops each roof. Each roof is then linked to the nearest businesses.

Classification relies on transfer learning: a pre-trained convolutional neural network extracts the features of each image and four dedicated classifiers recognise the structure (flat, shed, vault, gable…), the material (membrane, metal, fibre cement, asbestos…), the presence of photovoltaic panels and whether the roof is new. The classifiers are trained on roofs catalogued by the client’s technicians with a purpose-built web tool.

Models are trained and run on the company’s own servers, without sending images to external AI services.

In diagnostic centres there is not always a direct link between the exam booked in the schedule and the one admitted at the front desk: the same procedure is described in different ways (“cervical MRI” and “magnetic resonance of the cervical spine”), the patient turns up on a different day, or the admission is recorded without recalling the booking. This makes it hard to know for sure which patients did not show up.

We developed an agent that rebuilds these links automatically. For each booking–admission pair it compares the patient, the distance between dates and above all the procedures: descriptions are turned into semantic embeddings with a Sentence Transformers model and compared by cosine similarity, so that two different ways of writing the same exam are recognised as equivalent.

These signals feed a classification model trained on already verified matches, which assigns each pair a match probability and proposes the most likely admission for each booking. Bookings left without a match point to patients who did not show up.

The language model runs as a microservice on the facility’s own server, even without a GPU: patient data never leaves the company network.

Moving patients on stretchers inside hospitals means assigning, accepting and tracking missions in real time. For Formula Servizi we built a web platform for the control room and wards, together with a mobile app for operators: missions, parking bays and teams, QR code scanning of locations and offline operation.

The latest evolution strengthened security with two-factor authentication, encryption of sensitive data, an audit log and guided onboarding of Android devices via QR code.

Pharmaceutical and general supplies travel every day inside the hospital and to other facilities, often through intermediate transit points. For Formula Servizi we built a solution that manages transport units, cost centres, trips and deliveries, capturing requests from QR codes and barcodes.

Three dedicated interfaces, for the client, the control room and operators, and an app that also works offline, with digital signature, photos and temperature logs attached to each delivery.

For more than fifteen years we have been working with Lenti, a producer of cooked ham, in a plant where traceability is mandatory. The MES follows production from raw material acceptance to the start of production and the management of every step, through to the finished product.

Data is collected on the shop floor by operators or straight from the equipment: in several cases the system is connected to the machines, and where that is not enough we developed tailored Arduino-based solutions to talk to the PLCs.

Requests for medical aids and devices flow from local health authorities to suppliers according to tenders, contract codes and price lists, and the previous portal depended on a single management system. The new web platform, with separate front end and APIs, manages patients and districts, an ISO-classified catalogue and supplier prices.

It covers the whole cycle: prescription, authorisation, recurring orders, shipments, collections and sanitisation, with asynchronous integrations with suppliers’ systems.

In a plant of a multinational pharmaceutical company, shift and department handovers were done on paper. The responsive web app that replaced them offers configurable templates per department, conditional checks, work orders, email notifications and sign-in with company credentials.

In use for more than ten years, in 2026 it was integrated with SharePoint and its Oracle database was migrated to AWS.

Medicines and medical supplies delivered directly to patients, with temperature and priority constraints. The system follows every transport from request to delivery: planning, pickup, optional storage at the hub and signed delivery.

Clients and the control room work on the web, operators on a mobile app; delivery notes and temperature logs are attached to every delivery.

Aligned patient master data and reports published in the electronic clinical record, respecting the consents given. The solution integrates with the Master Patient Index through REST APIs, publishes reports using the HL7 standard with the related ADT messages and manages consent status.

Clinical and administrative documents to be collected, digitised and physically archived without ever losing track of them. For Formula Servizi we developed a web application that follows each document from collection to storage in the warehouse, including digitisation with photos and configurable metadata.

The platform exports data to public bodies and integrates with the systems of the hospitals involved.

Shipments and warehouse capacity to be organised starting from the plan generated in SAP. We built a decision-support web app that links hubs and destination countries, transport types, pallets and containers, with a shared work plan and notes for warehouse managers.

For Eliwell we developed software modules for the TelevisGo and TelevisBlue supervision platforms, used to monitor refrigeration systems in supermarkets and shopping centres. Components include the Modbus TCP interface for real-time data, the configurator, charts and reports, dictionaries and web services.

In a large international industrial group every new material code starts with a request: engineers and departments ask to create or change an item, the master data office checks it, asks for additions and approves it. We built the portal that governs this flow, with requests grouped in batches, ownership, shared notes and email notifications.

Each product family has its own technical attributes and dictionaries, from which the system builds descriptions in the different languages; search helps find existing materials before new ones are created. Approved data is written directly to SAP, and bulk imports from Excel speed up loading.

The Students to Business portal was built by Avarice for Microsoft to connect students, universities and Microsoft partners through internships. We managed it technically worldwide: the portal was active on 5 continents and in almost 100 countries, translated into 15 languages.

X-Icon Editor started as a project for Microsoft and then remained online as a free tool. It uses the HTML5 canvas to create high-resolution icons directly in the browser, exporting them in ICO format in four sizes: 64×64, 32×32, 24×24 and 16×16.

Try X-Icon Editor at xiconeditor.com

Inventorying thousands of assets spread across buildings, floors and departments means finding every item, recognising it, reading its labels and placing it in the right location. We built an app that, from a photo, reads the labels, suggests the asset type and finds similar items already recorded, all directly on the phone and even without a connection.

A web portal imports the initial list, guides locations with multi-level lists, tracks progress and produces the final file with derived fields calculated automatically.