Let’s Talk
Continue to explore
Oops! Something went wrong while submitting the form.
|---Module:text|Size:Small---| The evolution of Artificial Intelligence, Machine Learning and the use of real-time data is opening up new possibilities for business. At Celfocus, these technologies support solutions that make it possible to anticipate needs, trigger proactive actions and accelerate results in increasingly complex environments.
In an interview with Human Resources Portugal, Marco Ungari, Operations Intelligence Director at Celfocus, explains how the Cognitive Automation Academy, a programme that focuses on practical training and learning in real-world contexts to develop skills in Agentic AI, has been preparing employees for a new paradigm of work, where continuous learning and the ability to apply Artificial Intelligence to specific challenges are becoming increasingly relevant.
Here’s the full interview transcription:
Artificial Intelligence is transforming the labour market. Which skills do you believe will be most important in the coming years and what role should companies play in preparing people for them?
More than knowing how to programme a model, what will be crucial is the ability to design and orchestrate agentic systems, systems in which Artificial Intelligence carries out tasks autonomously, with human supervision at the right points, by breaking down a business process into tasks that an agent can perform, defining control points and validating results. This combines technical literacy with process thinking and critical thinking. Companies have a responsibility to create structured opportunities for this learning to take place in practice, including training in ethics and security in the use of Artificial Intelligence, rather than relying solely on occasional theoretical training.
What led Celfocus to create the Cognitive Automation Academy and what specific needs was this initiative designed to address?
The Cognitive Automation Academy was born out of a very specific need: the shortage in the market of professionals with practical, rather than merely theoretical, experience in agentic systems, Artificial Intelligence agents that, based on language models (the same technology behind conversational AI assistants), are able to understand a problem, decide on the steps needed to solve it and execute them autonomously. We needed a structured and replicable way of developing this experience in the field, alongside more experienced people, combining the reskilling of internal employees with the selective recruitment of specific profiles, with the aim of scaling this capability at a pace that recruitment alone could never achieve.
The academy combines the integration of young talent with the reskilling of internal employees. What advantages do you see in this approach and how do these two profiles complement each other?
Young talent have natural ease with the latest Artificial Intelligence tools and paradigms; internal employees bring in-depth knowledge of the business and our clients’ systems. By training them together, there is a positive ripple effect: more senior profiles gain technical fluency, and more junior profiles gain business context, and we create stronger mixed teams rather than training each group separately.
How was the programme designed to ensure that learning goes beyond the theoretical component and translates into skills that can be readily applied to the business?
The programme is structured in two phases: four weeks of hybrid training (online and in person), followed by a four-week practical phase on a real project, with close support from a senior mentor, during which participants develop agents that help solve specific project problems. At the end, each participant presents the results in a final presentation session, together with the mentor and the project sponsor, including a live demonstration of the agent and a clear explanation of the problem solved, what was done and the impact achieved, comparing the before and after.
Experimentation in a laboratory environment and the development of real projects are central elements of the initiative. What impact do these methodologies have on learning and the adoption of Artificial Intelligence?
Experimentation in a controlled environment allows people to fail fast and without risk, which is essential with technology that is still maturing. Real projects, on the other hand, give teams visibility of business value. As an example, we have already built agents that support the decision on whether to proceed with a release, assessing whether the accumulated risk of combining several deliveries into a single release is acceptable before it goes ahead, as well as agents that organise and cross-reference years of scattered documentation with the code itself, creating a single source of truth for teams. Seeing this type of tangible result helps unlock adoption, as it is easier to trust an Artificial Intelligence solution when you have seen the construction and validation process step by step.
One of the objectives of the Cognitive Automation Academy is to develop agents of change within the organisation. What role do you expect these employees to play in accelerating the company’s digital transformation?
We expect these employees to be able to act as bridges between technical teams and the business, helping to translate the potential of Agentic AI into specific use cases within their areas. This dissemination will also take place through more formal mechanisms that we are currently creating, such as a catalogue of reusable agents across projects and a portfolio of adoption initiatives, monitored at a dedicated management level.
Artificial Intelligence is evolving at a very rapid pace. How do you develop training programmes that are able to keep up with this speed of change?
By keeping the curriculum deliberately modular and reviewing it frequently, rather than treating it as a closed and static programme. We place greater emphasis on principles and ways of working, such as how to evaluate an agent, how to design control and security mechanisms and how to think about orchestration, rather than on a specific tool or model, because these principles age much more slowly than the underlying technology.
As this is a pilot project, which indicators will make it possible to assess its success and determine whether it should be expanded to other areas of the company?
We want to measure success through changes in behaviour across projects: the number of agents in production, the degree to which these agents are reused across teams, the impact on efficiency, for example, hours saved or response time, and the retention and progression of the participants themselves. The first edition included employees and projects from different areas of the organisation, and the development of agents is already being used in client projects. It is this evidence that will support the decision to expand the model to other areas.
What challenges have you encountered in implementing the Cognitive Automation Academy, both at an organisational level and in terms of employee engagement?
At an organisational level, the biggest challenge has been ensuring dedicated time for mentors, who fulfil this role alongside their project responsibilities, and ensuring that the academy is treated as a priority, with management sponsorship. More than a logistical issue, it is also a cultural shift, a willingness to change the traditional way of working.
What impact do you expect this initiative to have on attracting, developing and retaining talent, in a context where Artificial Intelligence skills are increasingly valued?
We believe this initiative will be a relevant differentiator, both in attracting young talent who want to grow in a cutting-edge technological field and in retaining experienced employees who want to continue developing without leaving behind the business knowledge they have already built. Investing in these skills in a structured way is also a clear sign that the company is committed to the development of its people in the medium term.
From your perspective, how should the relationship between reskilling, upskilling and continuous learning evolve as Artificial Intelligence becomes an integral part of organisations?
The distinction between reskilling and upskilling will become increasingly blurred, because almost every role will need to incorporate some degree of Artificial Intelligence literacy, even if it is not its core area. This means that continuous learning will no longer be a one-off investment at certain points in a career and will instead become a structural and permanent component of work.
What lessons from the Cognitive Automation Academy could serve as inspiration for other companies looking to prepare their people for the future of work?
The main lesson is that training in Agentic AI works best when, from day one, it is linked to real business problems, rather than treated as an isolated technical upskilling exercise. The second is that the role of mentors and visible management sponsorship are crucial, without this close support and the priority given by leadership, the learning curve would be much slower.
