Thanks to our DATA & AI Factory, we no longer show clients slides, but tools that are already up and running

Olivier Serfaty: That we were rewriting parts of the same tools for every project. With every new client, even if the end requirement was different, we’d rebuild the same frameworks, the same libraries, and the same connectors to SAP, Salesforce, ServiceNow or Microsoft Dynamics. It was a waste of time. So we created what we call ‘accelerators’ – ready-made building blocks that make up the ‘Gen AI Foundation’. When an Inetum consultant arrives at a client’s site today, they bring a ‘small suitcase’ with them. Inside that suitcase are these accelerators to help them get up to speed faster. In practical terms, this means around 30 per cent of a project has already been completed in advance.

Olivier Serfaty: There are two key factors. The first is these accelerators. The second is ‘vibe coding’: rather than writing every line of code, the engineer describes what they want to build in natural language, and the AI tools generate the code, which the human then validates or corrects. We save up to 40 per cent on development time. The direct result is that we can co-develop the solution with the client in ‘sprints’. We sit down with a ‘business’ representative. A week later, we come back with an initial version. They challenge it by giving us feedback: ‘That’s not quite right; I was thinking more along these lines ”. Another week on and we’ve met 80 per cent of the requirement. By the third delivery, we have the final product. Let’s remember that on traditional C++ or .NET projects, this was impossible: everything had to be specified before writing the first line of code. Today, we can get started, demonstrate something tangible quickly, and work towards a common goal together.

Olivier Serfaty: Demonstration units. For each of our four priority sectors (banking and insurance, the public sector, energy and retail), we already have proven use cases. When we identify a pain point for a client, there’s no point in giving a presentation; we bring a demo, tailored to their data where possible. A manufacturing company sees the tool running with their equipment requirements, a local authority with their urban planning data…

In the retail sector, for example, we’ve built a tool that analyses competitors’ websites and identifies ‘gaps’ in a client’s product range. When you have 250,000 product references, it’s impossible to do this manually. And AI goes beyond product listings: using just a simple photo from an online catalogue, it identifies a product’s characteristics (a corner sofa, a specific fabric), even where the text description is incomplete.

We create around three new use cases like this every month, and aim for around forty over the course of the year.

Olivier Serfaty: The Factory has three roles. To deliver quickly using accelerators. To win people over quickly using demonstrators. And to innovate constantly. We have a dedicated team that tests, benchmarks and experiments with new technologies. What we’re working towards is shortening the chain between the expressed need and the delivered tool even further. In future, during a meeting, we’ll describe the software we want verbally, and the AI will transcribe it, generate the specifications, write the code, and so on. We’re not quite there yet, but that’s the direction we’re heading in. The Factory is specifically designed to anticipate this moment, not to be caught off guard by it.