Brazil
Education
Implementing Organisation
Letrus
Brazil, São Paulo, São Paulo
Implementing Point of Contact
Thiago Pereira Rached
CEO
Contributor of the Impact Story
Carnegie India
Year of implementation
2017
Problem statement
Impact story details
Letrus is a pioneer and UNESCO award-winning edtech company that has been using artificial intelligence to improve students’ reading and writing skills at scale since 2017. Over its trajectory, Letrus has supported the literacy development of thousands of students and is currently implemented across states and municipalities throughout Brazil, working closely with public education systems to reduce learning gaps and promote equity. Our Literacy Program addresses these challenges through an AI-driven, evidence-based platform designed for students in middle and high school levels, that connects every layer of the education system, from students to ministries, into a single ecosystem for literacy improvement. Developed in Portuguese, Letrus combines advanced artificial intelligence techniques, including natural language processing, adaptive feedback models, and generative learning scaffolds. It supports reading comprehension and written expression, enabling equitable access to effective literacy instruction across diverse contexts. At the student level, the platform provides individualized real-time feedback on reading and writing tasks, helping learners identify comprehension gaps and refine expression step by step. It also enhances inclusive instruction by generating targeted insights into students’ learning profiles, including those with learning disabilities, language development delays, or from socioeconomically disadvantaged backgrounds, allowing differentiated pedagogical strategies within mainstream classrooms. At the teacher level, it saves time and improves instructional precision by showing class trends, common challenges, and recommended teaching strategies tailored to local curricula. At the school and district levels, Letrus offers actionable dashboards that inform principals and coordinators about literacy progress, supporting teacher development and data-based interventions. At the system level, education secretariats and ministries gain anonymized insights into regional trends, enabling equitable policy design and large-scale monitoring of learning outcomes. Letrus’ impact and rigor have been independently validated. A randomized evaluation conducted by J-PAL/MIT, one of the few global impact studies on an edtech tool, found significant improvements in writing proficiency among public-school students. Crucially, the study also showed a reduction in learning inequality, narrowing the gap between public and private school performance. This evidence confirms that Letrus not only raises literacy outcomes but also advances social equity.
AI Technology Used
Key Outcomes
Knowledge
Skills Impact, Inclusion
Equity, Efficiency
Productivity, Access
Reach, Accuracy
Quality Improvement
Narrative Outcome
Public school students in Brazil face a persistent challenge in writing proficiency. Teachers in under-resourced classrooms lack time to provide individualised feedback on student essays. Letrus uses AI to deliver detailed, personalised writing feedback at scale. The results show meaningful improvements in student essay scores. Students received feedback on significantly more essays, while teachers reported having more time for individual discussions rather than grading. In this way, the Letrus platform demonstrates how AI can complement teacher capacity rather than replace it.
Impact Metrics
Improvement in students’ writing proficiency (ENEM essay scores) through AI-driven feedback
Baseline Value
Average ENEM essay performance of public school students in comparison schools without access to AI-based writing support Standard deviations (SD)/ Exam score improvement
Post-Implementation
An increase of approximately 0.09 standard deviations in full ENEM essay scores among students was seen after using the AI platform Standard deviations (SD)/ Exam score improvement
Reduction of the public–private school achievement gap in writing
Baseline Value
Existing public–private school gap in ENEM essay scores prior to AI intervention Percentage of achievement gap reduced
Post-Implementation
A reduction of approximately 9% of the public–private school gap in essay scores was seen Percentage of achievement gap reduced
Frequency and volume of individualized writing feedback received by students
Baseline Value
Students in comparison schools received feedback on an average of 3.4 essays Number of essays with feedback/ Percentage increase
Post-Implementation
Students received feedback on approximately 1.3 additional essays (approximately 40% increase) and grades on 1.6–1.7 additional essays (approximate 45% increase) Number of essays with feedback/ Percentage increase
Reallocation of teacher time from grading to individualized pedagogical support
Baseline Value
23 % of teachers reported having “very insufficient” time to cover writing-related content
Post-Implementation
The share of teachers reporting “very insufficient” time fell to 9% among users of the AI-supported platform, while individual student–teacher discussions increased by approximately 35% Percentage of teachers/ Percentage increase in individual discussions
Implementation Context
States of Goiás, Mato Grosso, Espírito Santo, Piauí and municipalities of Rio de Janeiro and Joinville
Students in public middle and high schools, Youth from underserved and socio-economically disadvantaged communities, schools in urban and rural areas, schools with limited access to educational resources and connectivity, public-school teachers, school leaders and education authorities
Key Partnerships
Private and public education systems, education authorities, philanthropic foundations, research institutions, international organizations, Fundação Lemann, Fundação Bradesco, Abdul Latif Jameel Poverty Action Lab (J-PAL/MIT), Península Participações, Parceiros da Educação, B3 Social, and Imaginable Futures
Replicability & Adaptation
1. The implementation of the Letrus Literacy Program requires a combination of technical, human, and financial resources 2. Core AI technology is highly replicable 3. Effective adoption requires adaptation to local language, curriculum, assessment standards, and public education governance
Supporting Materials
* The data presented is self-reported by the respective organisations. Readers should consult the original sources for further details.