The adoption of generative AI at Repsol

With the launch of the second wave of our Digital Program (2023–2027), we took on a clear challenge: to lead the adoption of generative AI safely, responsibly, and with real impact on the business, keeping pace with digitization.

For this reason, and consciously assuming the role of early adopters, we launched the Generative AI Competency Center in June 2023, a trailblazing initiative in the European energy sector. The center's mission is twofold: to explore and develop the potential of this technology, and to do so always under a strong model of governance, ethics, and security.

What truly sets generative AI apart from other digital technologies lies in its ability to democratize AI. Any professional can interact, for the first time, with advanced models using natural language, thereby accelerating adoption and multiplying its impact on daily work.

To achieve our goals in accelerating this digital transformation, promoting new ways of working, and improving productivity, the competence center was initially deployed along four key lines of work, with a multidisciplinary team for one of them that integrates profiles from IT & Digital, Legal, People and Organization (P&O), and Audit and Compliance.

This line focuses on transforming the day-to-day life of our professionals, taking advantage of generative AI to free them from lower value-added tasks. Our approach is clear: AI does not replace people: rather, it enables new ways of working.

A four-month pilot with Copilot M365, in which 550 professionals participated, yielded very significant results:

• An average of 121 minutes saved per week, per person in task management.
• 16% increase in the quality of work performed.
• More than 60% of users expressed a preference for continuing using the tool, highlighting its positive impact on peace of mind, safety, and motivation.

This pilot confirmed that the true value of generative AI lies not only in saving time, but also in improving the work experience and empowering people. The study had such relevance that it led to a joint paper with MIT Technology Review, one of the first investigations worldwide on the impact of generative AI on daily work.

In our company's generative AI strategy, we call this type of personal productivity tools "gold dust." We have named it as such because it's very difficult to quantify and know what that those 120 minutes saved will translate into, where they'll go, and how they will be used used.

The second line focuses on identifying and implementing generative AI use cases that generate real value for the business by accelerating its digital transformation.

Here, the competency center lets us prioritize those cases with a direct impact on business processes. The competency center serves as a catalyst, promoting brainstorming sessions, designing initiatives and prioritizing investments, thereby significantly reducing costs and development times.​​​​

The indicators are clear: 

• More than 60 use cases in process across all our businesses 
• An investment of more than €2 billion. 

In this category we find our "gold nuggets": specific solutions with impact, such as the use of specialized models in the Legal area or the management of gas orders through WhatsApp. 

Generative AI has enormous potential to accelerate software development. During the 11th edition of HackIA, in collaboration with Microsoft, we experimented with GitHub Copilot, testing it in six key programming languages for our own digitalization efforts. 

The results were notable: 

• 200 participating developers. 
• Over 200,000 lines of code generated. 
• Reduction in development times and improvement of software quality. 

This laboratory confirms that generative AI is already a real accelerator of productivity in technological equipment.

AI adoption cannot happen without a solid foundation of ethics and accountability. From the start, we understood that scaling generative AI requires models that are secure, governed, and aligned with emerging regulatory frameworks. 

Our Responsible AI Group ensures the safe and responsible use of technology. We have defined six responsible principles, aligned with the European AI Act, and implemented a governance model that ensures compliance. 

In addition, we actively collaborate with external organizations such as IndesIA, OdiseIA, and the COTEC Foundation, reinforcing our commitment to responsible AI.

After a year of intense learning, new challenges faced, and lessons learned, in April 2024 we revisited our plan and identified the need to evolve our Competency Center with new lines of work. One of the key points was being able to identify truly transformative use cases, those that generated a tangible impact. To achieve this, we realized that we needed to systematize our methodologies. In order to identify, measure, and monitor the value that generative AI brought to different businesses, the Business and Productivity Impact Office was created. 

At the same time, we launched a technology observatory to develop reusable components and reinforce both data quality and governance. It also became especially important to focus on design and user experience: defining what the interface should look like and how we envisioned the interaction between people and technology in this new scenario. 

And finally, as a core pillar, we recognized the need to promote training programs that would help prepare and train all employee profiles across the company. Only with up-to-date knowledge and skills is it possible to guarantee a real and successful adoption of this new way of working. In fact, all of senior management and more than 5,000 employees have already been trained in prompting techniques for the effective use of these tools.

Just as with the first wave of our Digital Program, we're still looking ahead: today, to the evolution from generative AI to agentic AI. We decided to embrace this technology and, in addition to GenAI co-pilots and assistants with access to knowledge, agents capable of performing simple tasks now come into play. This also includes multi-agent systems, that is, systems capable of planning and executing complex tasks, making decisions autonomously, learning and collaborating. This last point is our "gold mine" since it is where we really manage to capture all the value of AI and obtain a return.

However, these agents need an ecosystem to coexist. Therefore, in collaboration with Accenture, we have developed our own platform for building intelligent agents. Currently, more than 34 agents have been deployed, working alongside 100 people to support critical processes such as requirements gathering, the solution development cycle, and incident management. The result is a complete reinvention of workflows, with a direct impact on productivity and operational efficiency.

The key to our digital transformation is not technology: it's people.

The collaboration between IT & Digital and the People and Organization area has been fundamental to helping our internal talent evolve into digitalized employees capable of applying advanced technologies and making decisions based on data.

We can highlight three key areas:

From the Repsol Data School in the first digital wave — with more than 4,000 students — to its integration into Repsol University's Digital Faculty, we have built a solid training ecosystem. Additionalt, there is transversal training in generative AI and prompting for all employees, with specific programs for managers and leaders. 

We have moved from an approach focused on the Digital Workplace to generative AI as a transformative pillar. Deploying co-pilots proves that change management — training, communication, and support — is critical to successful adoption.

We promote experimentation through initiatives​ ​​such as HackIA, a competition to accelerate digital use cases focused on AI, now in its 14th edition, and Hack It Yourself, where employees build their own apps, which has already empowered more than 4,000 makers to create 25,000 robots and 600 applications without the need for prior technical knowledge. 

The transformation we are experiencing is comparable to the internet revolution or the emergence of the iPhone. Generative AI is an accelerator of task reconfiguration and a technology that amplifies human capabilities. 

The value lies in reorganizing talent and allowing professionals to focus on activities that generate greater added value. The work evolves into mixed teams where humans and AI agents collaborate. 

My role and that of IT & Digital is to ensure that this technology is always there to serve people. By making a decisive commitment to generative AI and moving towards agentic AI, Repsol is not only a step ahead of its sector, but is also laying the foundations for a deep and sustainable transformation of work. 

Because when we talk about AI, we are talking about people.