Problem. Decision. Result.

The original application fell short in features, usability and technology, and was no longer developed. Documentation did not cover all processing rules, so part of the know-how stayed only with the people using the system.

At the end of the chain is an authority that checks the reports. An error in the data is not an inconvenience but a risk of penalties.

The shortcomings covered features, usability and the technology base. So we built a new application around users’ current needs.

It runs as a web application in the client’s Azure. We develop it incrementally, so benefits arrive before the whole, and modularly, so it can grow as reporting requirements change.

For some attributes, the column name alone was not enough to understand their meaning. Together with Nestlé’s people we gradually clarified what the values mean and how to calculate with them. This took the most time and decided whether the outputs were correct.

During the analysis we also found further manual steps - and automated them too.

The customer needed to fully replace an IAM product it had built itself. The existing solution was at the end of its life and the customer wanted a new platform on which it could expand its services.

The risk was not just technical ageing. A system handling identities and access is a security layer, and its failure or inability to evolve directly affects the customer’s clients, including international airlines, insurers and manufacturing plants, and the services that rely on it.

The constraint was the customer’s budget. There was no brief, requirements were defined on the go and development was iterative. It was also the customer’s first collaboration with an outsourced team from abroad; until then it had done everything in-house.

Such conditions shift the risk onto scope and budget: without acceptance criteria at the start, what counts as done has to be agreed continuously. Acceptance followed a simplified agile approach based on internal testing by the product manager, checking whether delivered features met expectations and the brief outline. The final acceptance criterion was the ability to roll features out to customers and feedback from end users.

We developed the platform to fulfil a requirement arising from a decision that was fully within the remit of the customer’s product manager and Architect Board. We used Keycloak as the foundation, with Spring Boot microservices on top and a React frontend.

As there was no brief, we helped the customer define the requirements. We delivered a complete project team: architect, analyst, frontend and backend developers, testers, a UX designer and a person responsible for documentation.

Our work also included regular reviews of architectural decisions, such as postponing the move to a Kubernetes architecture, which would otherwise have delayed the delivery of new features.