Colombia 🇨🇴 · Case Study
Mathematics for Everyone
Strengthening mathematics teaching in secondary education with AI and impact evidence
The Challenge
Nearly 40% of secondary students in Bogotá do not achieve fundamental mathematics competencies — a direct obstacle to the city's development and the social mobility of its young people. The Secretariat of Education set a clear goal: 70% of students will reach satisfactory performance by 2027. To achieve this, it selected Mentu as a strategic partner.
The Solution
An 8-month project with three integrated components across 58 public schools:
AI and Mathematical Mindsets Teacher Training
In-person bootcamps, virtual sessions, and implementation support, focused on the pedagogical integration of AI with Mathematical Mindsets (Stanford YouCubed) and Learning Variability (Digital Promise) approaches.
Contextualized Access to Shaia
Teachers had access to 5 specialized AI tools for mathematics, co-developed to integrate these pedagogical frameworks into classroom practice:
- Mathematical Experience: Design activities to develop mathematical mindset
- Pattern Conversation: Generate conversations that develop pattern recognition fluency
- Mathematical Worksheet: Design workshops that evidence student thinking and reasoning
- Student Feedback Practice: Enable teachers to practice mathematical conversations with an AI-simulated student and receive feedback on their facilitation skills
- Numerical Conversation: Find problems for numerical conversations that develop number sense and creativity
Rigorous Evaluation
Randomized Control Trial (RCT) led by IPA and independent researchers, with 270 teachers and 6,000 students evaluated.
Results
Improvement of 0.22–0.46 in mathematics results (RCT, IPA)
+6.83 percentage points in reasoning and problem-solving — the competencies with the greatest impact
The use of Shaia with integrated support generates significantly superior results compared to autonomous platform use.
Download the full case study
PDF report with methodology, results, and RCT evidence.