Artificial Intelligence
A 'smart copilot' to help you at work
A pioneering study reveals that, on average, each Repsol employee who used Copilot, Microsoft's smart assistant, saved 121 minutes a week in carrying out their tasks. This tool was particularly useful for three types of tasks: finding information, summarizing documents, and drawing conclusions.
In 2023, Microsoft introduced its 'smart assistant' Microsoft 365 Copilot, based on generative AI, which incorporates the power of its large language models (LLM) to applications such as Word, Excel, or Teams and adds as a knowledge context all the data that each company has in the 'cloud' and the information available on the Internet. This copilot offers real-time assistance according to the prompts, questions, or instructions formulated by the user, to facilitate the creation of content, perform data analysis, or synthesize documents. Copilot responds to each user taking into account only the information to which they have access within the organization, thus guaranteeing security and privacy in the corporate environment.
To test this technology, Repsol carried out a pioneering experiment with Microsoft 365 Copilot, the results of which were published in MIT Technology Review, the prestigious magazine from the Massachusetts Institute of Technology. This practical research involved 600 people, who received training before beginning a four-month trial period with this assistant. The aim was to measure how much time Copilot can save on daily tasks, to what extent Copilot improves the documents produced, and its effect on employees' work experience. The research also included experiments on real business situations, such as using Copilot to compare proposals from different suppliers or to prepare a presentation by compiling certain information.
One of the participants in this pilot program was Piedad Rodríguez, who has worked at Repsol for 28 years and works in the Global Services area, responsible for managing invoices for suppliers and employees. It was her first contact with this technology, which she thought "was something for younger people, from another generation, and that it would be difficult for me to learn, but nothing could be further from the truth." The first task she carried out with the assistant was to write an email for a supplier in which she had to touch on certain topics in a cautious way: "I told Copilot to help me write that type of email and it was a total success."
"The way you interact with it is very human-like, very intuitive," says Álvaro Guerrero, who works on the Voice of the Customer project, with which Repsol gauges consumer opinions to improve its services. "The only challenge is knowing how to give it the right prompts to convey exactly what you need. Once you get the hang of it, everything flows." Lorbada explains that "there is an inevitable learning curve. And once you cross that threshold, which Microsoft puts at just 11 weeks, you start to see the benefits and save time."
A 'smart copilot' to help you at work
Piedad, Álvaro, and Carmen have improved their daily work thanks to generative AI. They find this tool particularly useful when it comes to finding information, summarizing documents, and drawing conclusions from reports, allowing them to spend more time on value-added work.
Save time to spend it on more valuable tasks
The study reveals that, on average, each Repsol employee saves 121 minutes a week in carrying out their tasks thanks to this tool. Guerrero recalls his first test with Copilot as an example: "It was for a meeting that I couldn't attend, but that my colleagues recorded. I asked it to summarize the meeting and in a few seconds it had a perfect and detailed summary."
After the initial experience with this assistant, two out of three people at Repsol wouldn't like to go back to working without generative AI
Copilot was particularly useful for three types of tasks: finding information, summarizing documents, and drawing conclusions. Carmen Alabau, a sales manager in the company's Materials area, describes a use case: "In the sales department, we work with a lot of news to stay up-to-date with what's happening in our clients' markets, and Copilot produces some fairly high-quality summaries when you ask it to summarize a news item or a report. And if you ask it for information, it takes you directly to it, a task that can be quite tedious if you have to do it yourself, because you can't find it or it takes you a long time."
For more creative work, such as "focusing on a first draft of a presentation, you can also make a lot of use of the tool, asking it to propose ideas," Alabau continues. Something that Guerrero confirms: "It's the feature that has surprised me the most. It easily gives you suggestions when you are faced with a blank sheet of paper. You always have to revise its proposals, but they are a great help." The research showed that the documents presented by people who used Copilot were 16.2% higher in quality.
This set of improvements resulted in a very positive evaluation of the experience and two out of three people at Repsol would not like to go back to working without Copilot. Álvaro Guerrero sums it up like this: "Just as I wouldn't want to work nowadays without the Internet, I wouldn't want to work without generative AI either, because it's an ally that collaborates with you and makes your life easier."
"Generative AI is not a future opportunity, but a current one. At Repsol we are already taking great advantage of it and the challenge is to multiply that value as technology and its adoption advance," says Lorbada. The multi-energy company is deploying a very ambitious plan so that its more than 25,000 professionals have access to Microsoft's Copilot Chat assistant, a product that uses the same generative AI technology as Microsoft 365 Copilot. This smart assistant for the Internet "is changing how users interact with search engines, using natural language in their queries, and obtaining more elaborate and precise answers, without having to limit themselves to searching by keywords and all this in a secure, private, and confidential way, together with the possibility of creating agents or uploading files that can be used as a basis for prompts," highlights Antonio Cruz.
"AI will require us, as individuals and organizations, to train ourselves and reinvent how we work, and we must focus on the possibilities that these types of solutions open up for us, which come to complement us and multiply our capabilities," concludes Lorbada.
