80% of 10-year-olds in Latin America cannot understand a simple text. This statistic, documented by the World Bank and the OECD, has been cited in educational policy forums for over a decade. And yet, the problem persists. Why? Because we have faced it with solutions that prioritize speed over depth, and coverage over quality.

A crisis of decades, not days

The learning crisis in the region is not new. PISA, LLECE, and ERCE data show consistent trends: Latin American educational systems produce graduates who master procedures but lack deep understanding. It is not an individual failure of teachers; it is a systemic consequence of decades of fragmented policies, insufficient funding, and a huge gap between what is researched in universities and what happens in the classroom.

80% of 10-year-olds without basic reading comprehension. 60M+ Students at risk of falling behind in the region. 16 Countries where Mentu Labs operates.

The data is devastating, but not inevitable. The history of education shows that profound transformations are possible when political will, sustained investment, and methodological rigor are combined. Finland, Singapore, Estonia: the examples exist. The question is how to replicate those lessons in radically different contexts.

The trap of technological solutionism

Over the last decade, EdTech has promised to be the solution. Tablets, adaptive platforms, premium digital content, gamification apps… The reality is that many of these interventions have not moved the needle in terms of real learning, especially for the most vulnerable students. The problem is not technology itself; it is the premise that technology, on its own, can solve what is essentially a pedagogical and systemic problem.

“We cannot solve a pedagogical problem with a merely technological solution. AI needs committed teachers, and teachers need rigorous evidence that what they are doing works.” — Mentu Labs Team

Technological solutionism—the belief that technology can solve any complex problem—has spawned an industry of startups that measure success in ‘active users’ and ‘sessions completed’, not in real learning. This is not a moral judgment; it is a problem of incentives. When success is measured in usage metrics, it is optimized for usage. When it is measured in learning, it is optimized for learning.

Where educational AI does work

Teachers in the Dominican Republic exploring AI tools

Our experience in 16 countries has taught us that AI can be genuinely transformative when three conditions are met: it is integrated with quality teacher training, it is designed for specific contexts of use (not for generic markets), and it is rigorously evaluated with metrics that go beyond engagement.

The Lucero Project in the Dominican Republic is the clearest example: 250 educators co-designing 6 AI tools that responded to their real classroom challenges. These were not tools imposed from above; they were built from the questions that the teachers themselves asked. The result: a 40% increase in sustained use of the platform after 6 months, and measurable improvements in pedagogical planning.

Mentu Labs Principle: No tool is deployed without a rigorous MEL (Monitoring, Evaluation, and Learning) evaluation. The three levels—tool use, change in teaching practice, and student learning—are non-negotiable in any project we implement.

The three principles that guide our work

  • Evidence first: no tool is deployed without a rigorous MEL evaluation that measures real impact on learning, not just use.
  • Co-design with teachers: we listen, observe, and co-create before writing a single line of code. The teacher is the expert in their classroom.
  • Responsible scale: we prefer a deep impact in a thousand institutions than a superficial reach in ten thousand. Depth produces systems; superficiality produces statistics.

The road ahead

The learning crisis in Latin America requires political will, sustained investment, and methodological rigor. AI can accelerate progress, but only if we implement it with epistemological humility: recognizing that we do not know everything, that context matters more than the product, and that the teacher is the central agent of change in any educational transformation.

At Mentu Labs, we do not believe we have all the answers. We believe we have the right questions. And that, in an ecosystem dominated by superficial certainty and press optimism, is perhaps the most important thing we can offer.