Buenos Aires, April 2025. The main avenue is filled with thousands of protesters. Just next door, inside an AI forum, we were having a very different kind of conversation.

The soundtrack outside felt like a football stadium: drums, chants, and the occasional explosion in the distance. Our local hosts shrugged it off as business as usual.

It had become a race to inform a newly elected, inexperienced parliament, leading to a flood of regulatory proposals.

Inside, I was speaking with the policy director of one of Latin America’s most important employers’ organizations.

He was honest about the challenge his team was facing. Much of their recent work had become a race to inform a newly elected, and in some cases, inexperienced parliament, producing a flood of regulatory proposals. 

I'm seeing many proposals that are repetitive, poorly drafted, or even contradictory to existing rules. We agreed that it would be a dream to have an AI tool that could help his team do that, so he could focus on more impactful, forward-looking policy work.

That conversation stayed with me because it captured a challenge that is far from unique to one organization. Across countries and regions, policy directors from employers’ organizations have described the same structural problem.

They are expected to review, analyze, and comment on an ever-growing number of laws, decrees, and draft regulations, often with limited staff and little time. This is Mission Impossible.

ReguLens was born as an attempt to address that gap. The ACTEMP Team at the International Training Centre of the International Labour Organization developed the tool for and with the ILO's network of employer and business membership organizations.

From Design Thinking to Digital Glasses

The idea first took shape during a design thinking session with employers’ organizations from Panama, Costa Rica, and India. Together, we began sketching the earliest prototype of an AI-powered tool to help organizations analyze new policy and regulatory proposals more quickly and strategically. 

We called it ReguLens: a set of digital glasses for reading regulation.

We called it ReguLens: a set of digital glasses for reading regulation. The road from concept to usable tool was far from linear. I tested a wide range of prototypes with potential users. Some were too technical and assumed prior familiarity with AI tools, creating barriers from the start. That approach did not work. 

Over time, we learned that accessibility mattered as much as analytical sophistication. After several iterations, we moved toward a simpler model: users answer seven strategic questions to create a profile of their organization, upload a policy or regulatory proposal, and then choose from a set of practical functions.

Today, ReguLens can support users in several ways. It can assess the potential business impact of a proposal, identify contradictions with existing regulations, and help users think through possible advocacy strategies. It can also generate a curated map of relevant stakeholders, making it easier for organizations not only to understand a proposal but to respond to it more effectively.

AI as Enhancement, Not Replacement

What has surprised me most is not that the tool is useful, but how strongly users have responded to it. Several have told us that an analysis that takes around 60 seconds can save them days of work. For organizations operating under constant time pressure, that difference is not marginal. It can determine whether they enter a policy debate early enough to shape it, or only react after the fact.

For organizations under constant time pressure, that difference can determine whether they shape policy early or only react after the fact.

At the same time, one lesson has become very clear: this AI tool is not a substitute for human expertise, but as a force multiplier for it. ReguLens does not replace policy professionals. It helps them work faster, focus better, and spend more time on judgment, strategy, and advocacy rather than on first-round document triage. In that sense, human oversight is not a flaw in the ReguLens system, but one of its core strengths.

To that end, in order to access the tool, users must complete training to understand the tool's full potential and limitations. Training helps promote the responsible use of the platform and ensure that the AI serves to enhance rather than replace human judgment. 

Continuous Co-Creation

Over the past month, we gathered feedback through focus groups held in French, Portuguese, English, and Spanish, involving more than 250 users from every region of the world. What has been especially interesting is the degree of convergence in the feedback. Users from very different contexts often pointed to the same priorities.

One recurring request was comparative legislative intelligence. Users do not only want to know what a proposal says in their own country. They also want to understand how similar issues are being regulated elsewhere.

Whether the topic is working time, artificial intelligence, or another politically sensitive policy area, there is real value in seeing how peer countries have approached the same challenge. Comparative insight does not provide automatic answers, but it gives policymakers and stakeholders a stronger basis for informed debate.

That is why one of our ambitions for ReguLens 2.0 is to support cross-country legislative comparisons more systematically.

Building useful AI for the public good is fundamentally a collaborative exercise.

It has been one year since the Buenos Aires AI-Forum. If there is one broader lesson from this journey, it is that building useful AI for the public good is fundamentally a collaborative exercise. ReguLens improved not because a single team designed it in isolation, but because it absorbed ideas from colleagues across departments, from users across continents, and from repeated testing with people working directly on policy challenges.

The process has reminded me that designing AI is not only about technical capability. It is also about listening and adapting to your clients’ needs.

A final lesson is that AI is not an exclusive technology for those with the largest budgets or the most advanced technical teams. There is enormous space for practical, mission-driven tools that help organizations participate more effectively in policy debates.

ReguLens is one example of that possibility. It is still evolving, but it already shows that, when designed collaboratively and used responsibly, AI can strengthen the role of civil society in the policy debate.

Photo credit: © Adobe Stock / Supatman