Mexico’s digital government transformation is no longer a future ambition, it is already underway. In recent months, we have seen significant efforts to simplify administrative procedures, reduce processing times, and make interactions between citizens, businesses, and government easier. Data from Mexico’s Agency for Digital Transformation and Telecommunications (ATDT) reflects this progress: The number of procedures within the Federal Public Administration has been reduced by 49%.
These numbers matter because they show that digital transformation can have a very tangible impact on people’s lives and on how businesses operate. Fewer trips to government offices, shorter lines, less paperwork, and fewer opportunities for corruption.But as more government processes become digital, the challenge begins to shift. The next step is not simply to put more services online. It is to enable institutions, systems, and data to work together and, increasingly, to allow AI agents to interact with them without sacrificing security, traceability, or control.
After years of working in technology, I have learned that digitizing a process can be relatively straightforward compared with making it work as part of a broader ecosystem. Government systems were not built at the same time or according to the same standards. Each agency may have its own platforms, processes, and approach to managing information.
The real challenge begins when those systems need to work together.
No Need for a Single System
Mexico has multiple government agencies and different levels of government that have historically operated through separate processes and technology platforms. From a citizen’s perspective, those differences should matter less and less.
The National Pact to Eliminate Bureaucratic Procedures, signed by the governments of all 32 states and the Agency for Digital Transformation and Telecommunications, points precisely toward greater coordination across different levels of government.
This is an important step, but coordination alone is not enough. The real challenge is making different government systems work together without forcing every institution onto the same platform.
The goal should not be to build a single system for the entire government, but to make existing systems work together securely. That requires common rules for identity, access and data, as well as visibility into how information moves between institutions.
For citizens, if one government institution already has certain information, another should be able to access it securely when needed instead of asking for it again. A more connected government should ultimately mean a simpler experience for the people using its services.
AI Agents Change the Equation
Until now, much of digital security has been designed around people and applications. AI agents introduce a different kind of actor. Depending on how they are designed and the permissions they receive, they can retrieve information, interact with multiple systems, initiate processes, and execute actions.
This means governments need to consider not only what information an agent can access, but also on whose behalf it is acting, what it is authorized to do, and how far its autonomy extends.
An agent with excessive privileges could access sensitive information or perform actions it was never authorized to take. If something goes wrong, governments also need to be able to trace what the agent did, which systems it interacted with, and under whose authority it acted.
The challenge, then, is not simply connecting AI agents to government systems, but doing so with clear identities, defined permissions, and the ability to audit their actions. This is particularly important in government, where systems manage highly sensitive information, from identity and health records to tax and social security data.
As AI agents become part of digital government, integration and security can no longer be treated separately. Institutions need to be able to work together and adopt these technologies without losing control over the data and systems behind them.
The Italian Example
One experience that shaped my perspective on this challenge was a project developed with AlmavivA for Italy’s Ministry of Economy and Finance.
The ministry had multiple services and applications operating independently, with information distributed across different systems. The challenge was not to replace all of them or build a single platform from scratch. It was to enable those systems to work together.
The project introduced a centralized master data platform along with integration, authentication, authorization, and information governance capabilities. Today, it serves approximately 1.6 million users.
For me, the lesson is particularly relevant to Mexico. Interoperability does not mean that every government institution needs to operate on the same system. It means enabling different systems to share information and work together under common rules.
When critical information is involved, integration, security, and data governance cannot be treated as separate problems.
Digital Sovereignty and the Ability to Decide
To me, digital sovereignty does not mean developing every piece of technology domestically or isolating a country from the global technology ecosystem. It means maintaining control over how technology and data are used.
That includes knowing where data resides, who can access it, which systems and agents can interact with it, and what happens when a technology changes or fails.
As governments adopt AI, this ability to maintain control will become increasingly important. Mexico has already made significant progress in simplifying and digitizing public services. The next step is to ensure that institutions, systems, and AI agents can work together securely while remaining accountable and governed.
A more connected and intelligent government should not come at the expense of control. In the age of AI agents, the ability to govern how data and technology are used will become a fundamental part of digital sovereignty.