Teachers sense when a student is struggling but sometimes find out too late 

Colsubsidio is Colombia’s largest family compensation fund, a private non-profit organisation that has supported the wellbeing of Colombian workers and their families since 1957.

Today, the organisation serves around 1.5 million affiliates across the country, running its own network of schools (known locally as colegios) alongside health clinics, housing support and employment services.

Inside these colegios, teachers face a familiar but stubborn problem: spotting students who are falling behind before it is too late.

A teacher may juggle dozens of students across several subjects. For this reason, they may struggle to notice, from memory and scattered test results, which child has stopped keeping up.

There is however an existing tracking process. Or rather, existing processeS.

They consist in relying on manual observation and academic data spread across different systems (teachers’ informal notes on student behaviour, etc.).

In other words, a pattern a computer could catch in seconds can take teachers weeks to notice – if they notice it at all.

Without a smarter way to monitor learning, Colsubsidio risks leaving academic difficulties undetected for too long. 

This delays student support that works best when it comes early: 

• Extra 1:1 time with a teacher on a given weak skill,
• An informal end-of-the-class conversation to find out what’s actually going on (since a drop can be related to personal reasons),
• Extra practice material targeted at that one gap.

Neuroscience and AI, built for classrooms 

NeurekaLab is a Spanish startup that builds AI tools grounded in neuroscience to help children learn.

Its platform already supports schools, families and clinical professionals across Spain, working with children aged 5 to 12 on reading, counting, attention and memory.

For Colsubsidio, NeurekaLab adapted this platform into an AI-powered system that integrates data on students’ academic performance to build a live picture of each student’s progress.

By way of illustration, a teacher can open every Monday morning (or any moment they wish!) a dashboard with the following features:

1. Predictive analytics: 

• A teacher sees a student failing a subject mid-term – when there’s still time to act.
• School leadership can spot a whole class (or grade) falling behind, something only visible once individual flags are aggregated automatically.

2. Personalised learning paths: 

• A student who is weak in reading but strong in maths gets targeted reading exercises (not a generic progress track that includes irrelevant topics).
• Support can be recalibrated as the student improves.

3. Real-time data: 

A teacher checks the dashboard weekly rather than waiting for term-end grades, catching a drop within days rather than weeks or months.

Acceleration services that turned a trial into a measurable pilot

• Service n°1: PoC design & Prototyping

This service gives the partnership a structured framework for running the pilot at four Colsubsidio educational institutions. 

It sets clear KPIs around student engagement and measurable learning improvement, while defining the milestones for testing NeurekaLab’s platform with real Colombian classroom data.

“

This pilot project allowed us to validate, in a real-world setting, an artificial intelligence solution for personalised progress tracking of our 8,000 students in schools. More than just a technological test, it was an exercise that gave us evidence to drive educational innovation with greater confidence, for the benefit of millions of children in Colombia.

Juan Manuel BOTTÍA, Innovation Manager, COLSUBSIDIO

• Service n°2: Access to funding

This service provides NeurekaLab with expert guidance on navigating European Union, Latin American and Caribbean funding options to scale the pilot’s results.

It covers assessing which grants, loans, or equity funds the startup is eligible for, and advising on how to structure a funding strategy based on the evidence the pilot produced. 

The service cannot guarantee approval, but it provides the partnership with a clear map of where and how to apply to maximise the chances of success.

• Service n°3: IP assessment

This service reviews how the partnership should protect the intellectual property generated through the pilot. 

It clarified each partner’s contribution to the technology, and set out a plan for protecting, valorising and eventually transferring that IP across the countries where NeurekaLab hopes to expand. 

More than 70% accuracy for over 8,000 students

“

This pilot programme has been decisive: the tool produces the same results in the Colombian education system as it does in Spain, and it has shown us how to adapt it to the country's specific context to ensure its viability. We are ready to roll it out, confident in the positive impact it will have on students and in the classroom.

Sergi GRAU CARRIÓN, CEO, NEUREKALAB