from scientific question to working software

Now that code is cheap to generate, the value lies elsewhere: in understanding the domain, and in proving that the software does what it should. That is why we start with the scientific or business problem, the people who will use the solution and the rules it has to comply with. Only then do we choose the technology.
Our analysts, developers and AI engineers then build it in short cycles, together with the client's experts. The result is software that does what it needs to do, that users adopt, and that holds up under an audit.
We go into the details without losing sight of the bigger picture. And we keep learning, because the technology moves fast and our clients' fields do too.

how we work

AI makes it cheap to explore many options. What becomes scarce is the judgment to choose between them. That is why we split every project into phases where AI does the wide work, and phases where people decide.
Discover. AI helps us process interviews, documents, data and literature quickly. We look wide before we decide anything.
Define. We write the problem statement ourselves, and use AI to attack it. The outcome is a short spec: who it is for, what is explicitly out of scope, and success criteria we can test.
Develop. AI helps us generate and compare solutions fast. The success criteria from Define decide which one we build.
Deliver. People sign off, and in regulated settings we produce the validation evidence. What we learn in production feeds the next round of discovery.

01

05

Software development

We build software tailored to our clients' needs: from a web application for a research team to a platform that processes transactions in production. We work agile and in close collaboration with the users, and deliver documented software that the client can run and maintain independently. Every requirement can be traced to the decision behind it, the code that implements it and the test that verifies it. If needed, we stay on board afterwards for support, maintenance and further development.
architecture
back end
front end
data integration
cloud & on-premise
maintenance

02

05

Scientific business analysis

Our analysts have backgrounds in biomedical sciences, bioscience engineering and bioinformatics. They know the data types, the instruments and the regulatory requirements of the life sciences sector, such as data integrity in GxP environments. Because of that, the conversation with scientists starts from shared ground and requirements are right the first time.
domain expertise
requirements gathering
functional analysis
data integrity
story mapping
testing and user training

03

05

AI engineering

We design, build and run AI solutions ourselves: from machine learning on scientific data to agents that support a process. We experiment with local models and run our own internal AI setups, so sensitive data does not have to leave the house. Every solution is built so that a person keeps the final call and every result can be explained.
machine learning
agents
local models
human-in-the-loop
explainability

04

05

Design

The interface determines whether people actually use an application. We design for ease of use and clarity, from a first prototype to the final screens, and we visualise complex scientific data so that it becomes readable.
user research
UI/UX design
data visualisation
prototyping

05

05

Security & AI governance

We run an information security management system and an AI management system in line with ISO/IEC 27001:2022 and ISO/IEC 42001:2023. Every AI system we use or build is listed in an AI register, with a classification under the EU AI Act. For our clients this means that security and responsible AI are part of the process from day one, not added afterwards.
ISO 27001
ISO 42001
EU AI Act
privacy by design