How to introduce AI into the company: from processes to people

In the last twelve months we have met more than 200 companies and the reality is quite different from the one told online: the realities that have really integrated AI are still few, often held back by the fear of admitting that it is not as simple as it seems.

‍

AI in the company: an approach based on real experience

Introducing AI into the company today seems like the magic formula that everyone swears to possess, even though it often remains unobtainable. The truth that we have found, meeting more than 200 companies in the last year, is that there are still few realities capable of adopting it concretely; however, since admitting not using AI is perceived almost as a sign of backwardness, hardly anyone dares to talk about it openly. In this article, we will tell you some clear and objective points that we have seen help companies and that we continue to develop with our customers to transform innovation into real value.

‍

1. It's not a matter of technology, but of people

We should not think of AI simply as yet another software to add to our technological arsenal, but rather as a system capable of profoundly optimizing the work that we already carry out every day. Precisely for this reason, change can never start from the choice of technology itself, but must arise from the identification of the correct process and the right people who govern it. Once the workflow on which to intervene has been identified, artificial intelligence comes into play as an enabler that allows it to be improved or, in some cases, to completely upset it, with the sole objective of achieving the same business goal more effectively.

We are aware that this represents one of the most complex phases of the entire process, since it requires a synthesis between process skills and technological skills that often, unfortunately, reside in different departments or companies. Often, business leaders struggle to combine these two visions, leading to projects that remain confined to the test phase without ever generating a real impact.

Precisely to facilitate this encounter and overcome the initial barriers, we decided to make the journey to our laboratory completely free of charge. Our goal is to allow all companies to take this first step with us, offering a space to choose the ideal process, study how to modify it and test concretely whether artificial intelligence is actually able to achieve the desired result.

With this in mind, the best places to look for the first processes to be automated are in the backoffice or in any case processes that still have to do with many structured documents and a more or less standard process that seems repetitive to people and has low added value.

‍

‍

2. Choosing the right process (the 'Low Hanging Fruits')

Choosing the right process from which to start is not a trivial operation, but there are some fundamental parameters that can help us to tighten the circle and not waste energy. In our experience, there are mainly two things to keep under control. On the one hand, it is essential to select an environment managed by people who are really ready for change and who feel the need to solve a clear and well-defined problem. On the other hand, it is advisable to identify a process that is not overly complicated, but that is able to return a tangible and measurable result in a short time.

This second point serves to highlight a crucial aspect: as with any innovation, the hardest part is breaking the ice and getting started, but once you become familiar with the technology, taking the next steps becomes much more natural. To ensure the successful integration, we therefore suggest focusing on what are called “low hanging fruits”, i.e. those processes that are easier to fix that allow you to immediately validate the value of the investment.

If we were to give some practical advice on where to look, we would say to look for all those activities that have to do with managing significant volumes of textual documents, preferably typewritten. These are often repetitive processes that, however, still require a form of synthesis or human control to work properly. Very concrete examples that we see on a daily basis concern the automatic entry of data from documents such as DDT into management software, massive document analysis or various quality control checks. Generative artificial intelligence models are now extremely advanced and perform precisely in understanding the text, and it is for this reason that we often choose to start the business journey starting from these applications.

‍

3. The importance of privacy and data sovereignty

Even if we have emphasized that technology is not the only element that determines the success of an artificial intelligence project, it is still essential to know how to distinguish between providers that offer security guarantees and those that, on the other hand, must be viewed with some caution. In fact, infrastructure choices have a direct impact on data protection and privacy, two issues that today represent the main concerns for most companies.

As a general rule, we tend to consider the use of open source models preferable, since they guarantee much greater transparency on what actually happens to the data within the system. On the other hand, we must be realistic: it may be more complex to obtain from these models the same performance that characterizes closed source solutions. Furthermore, the technical challenge represented by the need to have dedicated hardware to run these architectures correctly should not be underestimated, a requirement that requires skills and investments that are not taken for granted.

Precisely for these reasons, we usually suggest collaborating with partners such as IBM and AWS. These are the two main providers that have strategically chosen to focus on secure platforms created to host and manage different models, including open source ones, instead of just promoting a single closed proprietary model. This approach makes it possible to build a much more robust perimeter of protection around business information, while offering the flexibility needed to scale applications securely and in compliance with regulations.

In conclusion, the integration of artificial intelligence should not be a laborious run-up to the latest news, but a multidisciplinary path made up of strategy, review of operating models and constant attention to the human factor. Success depends on the ability to move from isolated experiments to real and scalable production, always securing the company's data assets. Our laboratory remains an open and free point of reference for all those who wish to start this journey with the necessary support, transforming a complex technology into a concrete opportunity for growth.

‍

‍

Conclusion

From our experience, we have understood that the path of artificial intelligence is, first of all, a human revolution and not only a technological one. Finally, we have the opportunity to reimagine our processes and our companies, freeing ourselves from those old and rusty patterns that too often slow down everyday operations.

The adoption of AI should not be seen as a threat to our operations, but as an opportunity to return to putting people at the center of work, allowing them to focus on that process know-how that historically makes Italian excellence competitive in the world. Focusing on collective intelligence means using technology as an enhancer of human capacities, not as a substitute for them.

Precisely for this reason, our laboratory remains open and free: we want to offer companies a physical and mental place where they can take the first step without the economic risks of investing in the dark. It is a concrete opportunity to discuss, test your ideas and define together a roadmap that is, finally, realistic and sustainable for your reality.

‍