The adoption of artificial intelligence is advancing rapidly among Spanish companies. By 2025, 20.5% of companies with ten or more employees were already using AI technologies, a share that reached 57.5% among large enterprises, according to the National Observatory of Technology and Society.
However, for Docenteo, the ease of developing pilots and early solutions has highlighted that it is much easier to experiment with artificial intelligence than turning those initiatives into stable tools with real business impact. In an environment where technology, competitors, and needs are changing ever more quickly, it is no longer enough to wait until the end of an AI project to verify whether it is delivering value: the vision may be long-term, but the impact must arrive sooner.
“The objective should not be to do more AI projects, but to make companies more competitive by changing how they operate with AI. And that change only makes sense if it ends up translating into an impact on the bottom line,” says Alejandro Peris, cofounder and head of business at Docenteo.
Based on its experience with AI projects, data, and technology transformation, the company identifies five key practices for an AI initiative to move beyond the experimental phase, deliver results, and increase the organization’s ability to tackle new challenges.
Start with impact, not with technology
The starting point should not be the technology, but the result you want to achieve with it. For Docenteo, any project must begin with a concrete metric to move and a goal that allows you to determine whether the initiative has succeeded. Reducing the cost of a process, boosting productivity, or improving the experience only counts if there is a baseline to compare the outcome against.
Without knowing where you stand and where you want to go, it’s impossible to know if anything has been improved. AI should not become an end in itself or a race to accumulate use cases. It should be a means to generate a measurable improvement, whether in growth or efficiency.
Pursue tightly scoped, measurable, and scalable projects
In the face of large, closed deployments, Docenteo recommends developing initiatives that allow testing a hypothesis, delivering results quickly, and adjusting the approach when necessary. In a world that’s changing faster all the time, waiting months to determine whether an initiative creates value increases the risk of arriving late.
The key is not to curb ambition but to break it into more concrete projects that demonstrate impact, accelerate learning, and build capabilities that can scale across the organization.
Design with production in mind
A pilot can tolerate a lot. However, most never make it to production. The real challenge is getting them to work at scale within the organization. Today, the major AI challenge is precisely industrializing the use cases: changing processes, data quality, integrations, operating costs, permissions, security, traceability, or maintenance.
Therefore, these elements must be built into the design from the start and not treated as a later phase.
Blend business and technical know-how
People who work daily with a process know its problems and opportunities, while technology specialists bring insight into AI’s capabilities and possibilities. For Docenteo, the best results don’t come from a lab: they appear when both perspectives work together on the ground, from identifying the use case to its implementation.
This collaboration should also serve to transfer knowledge. A project should not be limited to delivering a solution but should leave the organization with judgment, practices, and the capacity to operate and evolve what has been built.
Build the capacity to keep evolving
Putting a solution into production is not the end of the project; it’s the moment when change begins. It’s when the solution must integrate into processes, become part of how teams work, and turn into a new organizational capability. For Docenteo, deploying AI also means transferring the knowledge and the criteria necessary for people to operate and evolve what’s been built.
Thus, success is not measured only by what is delivered at the end of a project, but also by what the organization is capable of doing once the vendor is no longer involved: evolving the solution, applying what’s learned, and continuing to advance with greater autonomy and internal capacity.
From pilot to real impact
For the company, the next stage of artificial intelligence will be defined less by the number of tools or pilots developed and more by the organization’s ability to integrate the technology into its processes and evolve it as needs change.
In this scenario, relying on specialists will remain necessary, but companies cannot outsource their ability to change. The role of technology partners must go beyond solving the current project: they should help the organization be better prepared to tackle the next one.
“The success of an AI project should not be about leaving a solution running. It should be about leaving a company better able to understand the technology, evolve it, and face the next change with less external dependency. That capability is what ultimately turns AI into a real competitive advantage,” says Álvaro Montero, cofounder and chief technology officer of Docenteo.
