We respect your privacy
We use cookies to keep the site running and collect optional metrics. You can review your choices at any time.
View Cookie PolicyReflections
Short reflections on digital transformation, enterprise architecture and project management.
We know how many people use a public service, but almost never whether it solved their problem - and without that measurement there is no way to tell whether the AI now being pushed everywhere brings value. The proposal is to flip the order: use AI itself to build the measurement we never had, before using it to automate.
Starting from Anthropic's study on AI's impact on the labour market, this reflection explores what happens when we apply that logic to Portuguese public administration: the paradox between a shortage of people and a surplus of automatable roles, the problem of attraction and exit, and why the current model does not work.
Before replacing a system that has been running for 15 or 20 years, it is worth understanding what it solved. This reflection explores the risk of undervaluing legacy, the importance of listening to those who maintained it, and the tension between respecting what exists and knowing when to cut.
Introducing AI to speed up public procurement may make the problem more visible, but it does not solve it. The biggest friction lies in what gets contracted, how success is defined, and how delivered value is measured.
AI makes “build” a practical option in the public sector by reducing friction and dependency, while demanding stronger architecture, governance, and technical accountability.
Digitising public processes without simplifying them only computerises old problems. The real challenge is keeping a simple path for most users while giving dignified routes to exceptions, without killing the MVP or betraying universality.