There is a difference between talking about artificial intelligence and working with it. At CENCO we chose the second, and we chose it a long time ago: not as a marketing layer on top of what we do, but as part of our technical team’s daily process. Both to work better internally and to build better products for the outside world.
What changes when an AI assistant knows your code, not just the language?
When someone on the team needs to estimate a project or understand a system they didn’t build, they don’t start from scratch or ask “by eye”. We use an assistant trained to navigate large codebases and point out, with concrete evidence, which file, class or method holds the answer to a technical question. It’s not a tool that offers opinions about the code: it’s one that walks through it and tells you exactly where to look.
We use the same approach in support and maintenance, not just in development. A real, day-to-day example: a support colleague needed to answer whether the messaging system could send campaigns only to a specific user role. The first internal answer was ambiguous —an “I think so” that isn’t good enough when someone is about to run that campaign in production—. When we asked the assistant, the answer not only confirmed it was possible: it added the nuance the first answer had left out. The generated filter had to be reviewed before activating it, and we had to confirm that users had the corresponding role correctly loaded in the database, because otherwise the send simply wouldn’t reach them.
That difference —between “yes, it can be done” and “yes, it can be done, but check this before you activate it”— is exactly what separates a fast answer from a correct one. And it’s the reason we integrate these assistants into tasks where a mistake is not a minor detail: it’s a support ticket that turns into a production incident.
What if the internal tool becomes the product?
That same principle —using AI agents as part of the work process, not as an add-on— is the foundation of our own agent platform. It started as a lightweight tool, designed to orchestrate several specialised agents on top of state-of-the-art reasoning models. Over time it became a product with a real interface: organised projects, conversation history and different working modes depending on what’s needed —querying data and getting results in exportable tables, querying code or generating automatic summaries of an entire project—.
We built it for ourselves first. That means when we bring it to a client, it’s not a demo put together for the occasion: it’s a tool we already use in production, every day, for our own work.
What does this look like applied to a real business?
That same approach translates into concrete solutions. One case: a support product for salespeople in the retail channel, designed as a conversational assistant that combines three different sources of information: the company’s internal documentation, the database with each salesperson’s own history and sales, and the public catalogue with products and prices. The salesperson asks in natural language, and the agent cross-references those three sources to give them a useful answer in the moment, on the shop floor: not a generic answer, but one built with their own numbers.
It’s not a chatbot connected to an FAQ database. It’s the same logic of combining disparate sources of information that we use internally, applied to a concrete business problem.
The underlying idea
We don’t believe you have to choose between talking about AI and using it. At CENCO the order is the other way around: first we build it into how we work, and what we offer comes from there. That’s why we can propose agent solutions with confidence, because before proposing them to a client we have already tested them on ourselves, in our own day-to-day work.
How can CENCO help you?
At CENCO we have been integrating AI agents into our own work and our clients’ for some time. We don’t build demos: we build tools that work in production, tested internally first. If you want to understand how this approach could apply to your team or your industry, we can help you take the first step.