Miguel Ángel Viedma

Companies invest millions in Artificial Intelligence licences, yet most projects fail not because of the technology, but because of poor adoption. These are the key steps for bridging the gap between having AI and actually working differently because of it. A large company proudly announces that it has deployed Artificial Intelligence: 100% of its workforce now has a licence for a cutting-edge AI assistant. Six months later, most employees use it as a slightly more articulate search engine, many have never opened it and hardly anyone has changed the way they work. One hundred per cent licence coverage, but real adoption remains limited. This increasingly common scenario captures one of the biggest misunderstandings of our time: confusing having AI with adopting it. The problem begins with the rush to keep up. Fearful of falling behind, many organisations race to buy licences without first asking what they actually need them for. But every change involves a process: trying to accelerate it is like learning to ride a bike without first using stabilisers. The goal is perfectly valid, but the shortcut often ends in frustration. The conclusion is as uncomfortable as it is encouraging: AI projects often fail not because of the technology, but because of poor adoption. And adoption does not begin on the day of technical deployment, but much earlier: with how we prepare people to work differently. Below, we explore five key areas for anticipating the most common mistakes and turning AI into a capability that is genuinely embedded in everyday work. AI is not a technology project, it is an organisational transformation Implementing a tool does not guarantee that people will use it, and using it does not guarantee that they will get value from it. The aim of AI is not simply to help us do the same things with a different click: it is to change how we work, collaborate and make decisions, through sustained behavioural change and new habits that endure, rather than a two-week burst of curiosity. For change to take hold, it needs to spread from within, and this is where a network of AI champions can make all the difference. A common mistake is choosing the most technically skilled person to lead it instead of looking for the most influential person, the one everyone listens to even if their influence does not appear on the organisation chart. Change does not spread by decree, but through a little healthy envy: the question that gets an organisation moving does not come from a committee, but from a colleague who sees the results and asks, 'How did you finish that report so quickly?' To lead change, do not look for the most technical person, but the most influential. AI adoption spreads through healthy envy, not by decree. From doing to supervising: how people's roles are changing When AI is adopted effectively, it changes the role of the person using it. The shift is from an execution mindset to one of supervision and co-creation. Instead of spending time starting from scratch (the first draft, the endless summary, etc.), professionals can focus on validating and improving what the AI produces, applying the expert judgement that AI does not have. It helps to think of AI as a brilliant but inexperienced junior assistant: fast and productive, but the value lies not in delegating without checking, but in supervising effectively. Although this shift frees up time for what really matters, one of the most costly mistakes is never far away: believing that simply adding a layer of AI to what we already do is enough. Value does not come from adding a tool to an existing process, but from redesigning the process itself. Automating an inefficient process only makes it inefficient faster. The greatest impact comes not from speeding up what we already do, but from redesigning it. All of this requires new capabilities: not so much knowing which button to press, but learning to ask good questions, challenge the answers and combine human judgement with the power of the technology. Leadership is not announced, it is demonstrated There is a less visible mistake that undermines many initiatives: confusing sponsorship with leadership. Leadership means setting an example: funding the project is not enough; leaders need to use the tool themselves and model the expected behaviour. In some companies, including consultancies, effective use of AI is already beginning to form part of the criteria considered in career progression. It is a sign that AI adoption is becoming less of an optional choice. An email from the CEO announcing the company's commitment to AI changes nothing. What changes an organisation is seeing its leaders use it. There is another group that is often overlooked: middle managers. They translate strategy into day-to-day practice and give people explicit or implicit permission to learn and make mistakes. Middle managers who believe in the change and actively support it can be one of the strongest predictors of success; sceptical middle managers can be one of the biggest barriers. Cybersecurity AI & Data Security and compliance in Generative AI applications like ChatGPT and DeepSeek April 15, 2026 Putting people at the heart of the transformation No transformation of this scale can succeed if it treats people as an obstacle rather than placing them at the heart of the process. In this respect, four ideas are crucial. Building capabilities takes more than an afternoon of training. A one-off training session will not transform an organisation. Capabilities are built through practice, repetition and ongoing support. Forcing adoption breeds resistance. Making people use AI creates resistance and minimal engagement; genuine adoption comes from trust and from each person understanding what they stand to gain. Not everyone starts from the same place. Every organisation includes people who are already experimenting on their own and others who find a blank prompt box daunting. Treating everyone the same leaves some bored and others paralysed. Managing fear is part of the job. Much of the resistance is driven not by laziness, but by anxiety: fear of becoming obsolete or that AI is coming for my job. The answer is to communicate honestly and reframe the narrative: when properly integrated, AI can free up time so that people can focus on higher-value tasks. Adoption cannot be imposed, it has to be earned: it grows from trust and from each person understanding what they stand to gain from using AI. What is neither communicated nor measured will not be adopted An invisible transformation might as well not exist: however good the results are, if nobody sees them, they will not build momentum. Consistent communication turns the curious into users and users into advocates. Success spreads when it is shared. But communicating well requires measuring well, and this is where the more subtle mistake lies: measuring what is easy rather than what creates value. Counting licences or queries says nothing about impact. The real question is what changes as a result of using AI. The biggest measurement mistake: counting active licences is easy; quantifying the value generated is what matters. The two rarely align. And what gets recognised gets repeated. Recognition —celebrating someone who finds an ingenious use for AI and bringing it into performance conversations— is one of the most powerful and cost-effective drivers of adoption. It is not simply the result that is rewarded, but the willingness to explore, showing that using AI effectively can move from being an obligation to something people aspire to. From 'having AI' to 'working differently because of AI' Ultimately, it all comes down to one fundamental distinction. The goal was never simply to have AI, but to become a company that works differently because of AI: one that makes better decisions, focuses its talent on what matters and has made technology an embedded capability rather than a decorative feature in the annual report. That difference, between buying and transforming, separates companies that talk about AI from those that genuinely benefit from it. The gap is not bridged by buying more licences, but through greater adoption: leaders who set the example, people at the centre, redesigned processes and a culture that embraces change rather than fears it. AI does not reward the companies that buy the most technology, but those that are best at changing the way they work. In many cases, by this stage, the technology itself is rarely the problem. The challenge and the opportunity lie with people and their willingness to work differently. Organisations that understand this sooner will not only have better tools, but will become different organisations altogether. ■ At Telefónica Tech we help companies adopt AI effectively by combining technological capabilities, professional services and specialist expertise to embed it into the way they work, improve efficiency and generate business value. Find out more → ______ AI & Data Data Governance: a great ally to put limits to Artificial Intelligence October 31, 2023