The reading fluency assessment that brought evidence of learning to public school systems nationwide

Saber Educação is the K–12 education division of Grupo Cogna Educação, a publicly traded company on Brazil's B3 exchange (ticker COGN3). For decades, it has been a leading provider of educational content and teaching systems in Brazil, operating nationally recognized brands including Saraiva, Ática and Scipione. Its materials reach hundreds of public and private school systems across Brazil, consolidating its leadership in K–12 education. It is recognized for the academic quality of its content and continuous innovation in educational solutions. Throughout its history, it has transformed how millions of Brazilian children learn.

The Challenge

Carefully assessing each student's reading fluency has always been demanding manual work, requiring time and the expertise of an attentive evaluator. Even within one public school system, it is a logistical challenge. Multiplied across the dozens of systems using Saber materials throughout Brazil, each with different infrastructure and capabilities, the problem becomes one of method as well as resources. No manual operation can sustain that volume.

Saber needed both depth and reach. The more careful and individualized a reading assessment is, the more it demands of the evaluator. And the more the solution must scale, from a municipal system to an entire state network, the less room remains for that individual attention unless technology takes on part of the work.

The Solution

Saber retained full educational authority over the product vision, deciding the curriculum, fluency criteria and how each report should guide classroom planning, aligned with Brazil's BNCC, PNA and PNE. Radhark took responsibility for the end-to-end technical architecture of Sab.IA, from capturing reading aloud to generating final reports for teachers, schools and school systems.

The solution automated both the administration and analysis of reading aloud using generative AI. Students complete three timed one-minute readings of increasing difficulty: simple words, uncommon words and a narrative text, all recorded. The AI model analyzes the reading, calculates correct and incorrect responses by item and overall accuracy, and organizes results into reports by student, class, school and school system. The platform works on computers, tablets and smartphones, allowing each system to assess students using its existing infrastructure. It runs entirely on Google Cloud, using the same generative AI foundation as Radhark's other reading fluency solutions.

The Result

Saber gained a way to verify, school system by school system, whether the materials it distributes nationwide are actually helping children learn to read. Assessment moved from a time-consuming manual process to an automated one that works on devices the schools already have, delivering consolidated reports by student, class, school and system within the same school year.

This transformation is important to Saber's commercial proposition. Previously it offered educational materials backed by publishing history and brand reputation. Now it can offer materials with real-time evidence that children are learning from them. For public school systems that must demonstrate learning outcomes to administrators, this is the difference between a promise and proof—one that can influence contracts and renewals.