By Javier Aguado, Chief Delivery Officer at ALIANDO
Currently, there is a paradox in the business world: while the vast majority of companies are adopting Artificial Intelligence, very few are achieving the expected results.
Since the public launch of ChatGPT in late 2022, the wave of adoption of Generative Artificial Intelligence has been advancing unstoppably in the business world. Whether driven by its practical applications in specific business processes, initiatives from the innovation department, pressure from senior management, pressure from shareholders (who see how any AI-related news positively impacts the stock price), or even the curiosity and demand of employees themselves.
The figures speak for themselves: 81% of leaders expect to include AI agents in theirstrategies¹, and 93% of organizations are already experimenting with variousAI models².
However, while some companies are achieving significant improvements in EBITDA and are able to roll out applications in less than three months, most are facing serious difficulties in realizing a return on their AI investments: more than 90% of companies are not seeing significant financial returns; and despite the time invested, fewer than one-third of projects move beyond the proof-of-concept stage.³
This paradox surrounding the use of Artificial Intelligence presents a dilemma for anyone responsible for promoting or leading these types of projects. On the one hand, the market is inexorably pushing for adoption; on the other, the figures indicate that it is necessary to be much more cautious with investments in Artificial Intelligence.
After more than 10 years of working with hundredsof clients on initiatives related to Artificial Intelligence—both generative and more traditional (Machine Learning, Deep Learning, etc.)—,and ALIANDO have identified the three main causes of failure in the adoption of Artificial Intelligence in business environments:
Incorrect selection of use cases
The biggest mistake many companies make begins right from the start, when they decide where to implement AI. A fundamental error is to start adopting AI without calculating the ROI (return on investment), without validating whether users will adopt the tool or whether the technology should be used to solve that particular problem. Not all challenges require AI-based solutions; there are other technologies (RPA, automation, advanced analytics, etc.) that can offer a more effective solution. Therefore, it is essential to have an“envisioning”methodologythat identifies, categorizes, and prioritizes AI use cases. Choosing the wrong use cases means investing in flashy but irrelevant pilot projects; when budget reviews come around, those projects will be seen as “expensive toys.”
Lack of Governance
As with any other technology, it is essential to establish a governance model that, through policies, teams, and tools, sets the framework for the lifecycle of AI projects. In the case of Centers of Excellence (CoEs), there is no need to create large, complex structures; rather, a pragmatic approach should be taken to define the boundaries and procedures necessary to operate AI safely and efficiently.
Without a clear governance model, security, compliance, and reputational risks skyrocket, and any incident associated with AI will quickly lead to a search for someone to hold accountable.
Lack of Context
As with any analytics-related project, the data provided to AI tools is essential to its success. A “Context Engineering” practice that ensures the data fed into the AI is appropriate and accurate is essential for the application to function properly. Likewise, it is necessary to define the mechanisms for managing this data, its validity, and the associated responsibilities. Without proper context, AI not only makes mistakes—it does so with such confidence that it can mislead users and executives, eroding trust in the technology and those behind it.
Ignoring these three aspects not only multiplies the likelihood of AI failure but also increases the risk that the investments made will not generate the expected return.
Conversely, implementing these three lines of work—coordinated and under the umbrella of a corporate AI adoption strategy—significantly increases the likelihood of deploying AI efficiently, securely, and sustainably over the long term. In this way, AI can become a real advantage for the company—rather than the reason why someone might have to look for a new job.
Sources
1. Work Trend Index Annual Report
2. 16 Changes to the Way Enterprises Are Building and Buying Generative AI | Andreessen Horowitz
3. McKinsey, 2025; BCG, 2024; Adecco, 2025; Gartner, 2023.
