An AI Master Plan is not a tool catalog or a list of pilots. It is an institutional decision on how an organization wants to understand, prioritize, govern and review the use of AI. The Project of theTerrassa City Council You can explain this process from the inside.
This piece adopts the format of editorial conversation with Ignasi Llorente from the materials and documented results of the project. It is not the literal transcription of an interview nor does it attribute non-validated textual statements. The objective is to order the questions that guided the work and the answers that the case allows to sustain.
Why does a city need an AI Master Plan?
Because AI usually gets in before the strategy. People provide assistants, automate tasks, generate content or consult documents from different areas. This move can bring value, but it also creates criteria inequalities, data risks, duplication and incompatible decisions.
The Master Plan turns this scattered reality into a shared policy. It defines what you want to achieve, what principles cannot be violated, who decides, how use cases are evaluated and what capabilities the organization needs. In a public administration, in addition, each decision must be aligned with the rights of citizenship, transparency and the general interest.
What was your first job in Terrassa?
Listen before prescribing. The diagnosis combined an internal survey with 341 answers, spaces of participation with 160A webinar with 63 participants and an idea with 48. This set allowed us to observe real uses, expectations, fears, frictions and opportunities from different points of view.
The important data is not just the volume. It is that the Plan was not built from a single office or from a technological demonstration. People who know the municipal procedures provided the necessary context to distinguish an attractive idea from a real public need.
How do you make hundreds of contributions to a strategy?
First you have to order. Contributions were grouped by needs, areas, potential impact, dependencies, risks and maturity. They were then contrasted with the principles of governance and the actual ability of the organization to deploy them.
The process identified 25 proposals pending prioritization. We do not present them as implanted pilots. These are opportunities that need a subsequent decision on public value, viability, data, responsibility, cost and supervision. This distinction protects the project from rushing to announce results before they have been validated.
What does a Master Plan contain that can be applied?
The final document was structured in 14 chapters and a horizon 2026–2028. It integrates strategy, governance, ethics, regulation, literacy, prioritization, monitoring and cultural change. The result does not separate technology and organization: it shows which decisions depend on data, people, processes or government bodies.
It also proposes an architecture of responsibility. Strategic spaces define priorities and validate sensitive uses; the operational structure coordinates projects and maintains the registry; the referents of the areas connect the Plan with the daily reality. Governance is not a committee that appears at the end: it is the system that accompanies the entire life cycle.
Why is participation also a tool for governance?
Because it expands the quality of the decision. A legal area can detect a risk that is not visible to technology. A person of citizen attention can explain an exception that the theoretical process does not collect. A data team can identify that a proposal does not yet have the necessary information.
Participating does not mean that all ideas are executed. It means that priorities can be explained, that boundaries are visible and that people understand how to be part of the change. This legitimacy is especially important when the AI may affect services, rights or public decisions.
What does Terrassa teach other organizations?
That starting well is more important than starting with a spectacular tool. In five months we were able to build a diagnosis, activate spaces for participation, organize proposals and draft an institutional framework. The value was not to promise immediate automation, but to create the conditions for future projects to be comparable, governable and reviewable.
The Municipal Assembly approved the Plan in April of 2026. The Official information of the City of Terrassa ↗ It confirms this institutional step. From here begins a less visible and more decisive phase: turning the roadmap into capacity, keeping the log alive and reviewing what works.
The Question That Should Be Opened
A good Director Plan does not close the conversation about the AI. Set how to continue it. Each new use case must be able to answer what it contributes, what sources it uses, who assumes responsibility for it, how it is tested and under what conditions it must be stopped.
This is the criteria that we transfer from Terrassa to other administrations and organizations: Governing before climbing doesn’t stop innovation; it gives it a direction that can be sustained.