Research

Working together, our world class and interdisciplinary research teams have spent over a decade understanding the why behind students’ struggles with learning.

Awarded for Evidence

Eedi has earned a gold certification for Efficacy from EDUEvidence for demonstrating statistically significant impact through randomised controlled trials (RCTs). These RCTs were independently verified by What Works Education, a UK-based evaluator.

Our Research Teams

Our research is driven by three collaborative, interdisciplinary teams.

Evidence ->

Our Evidence team measures real-world impact. It designs and runs rapid tests and gold-standard Randomised Controlled Trials (RCTs), independently validated, to demonstrate real, measurable improvements in learning outcomes for students.

Data Science ->

Our Data Science team builds the sophisticated models that power our Diagnostic Engine. They translate academic research and real-world data into the predictive, practical AI that underpins our product ecosystem - including end-user products like Eedi Tutor and DQR.

Learning Design ->

Our Learning Design team of experienced former teachers is responsible for the integrity of Eedi's diagnostic questions and misconception library, ensuring every question, construct, distractor and explanation is age-appropriate and evidence-informed.

Research Principles

Four core principles guide our approach to building the future of learning.

Grounded in the Classroom Reality

Eedi was built by teachers, for teachers. Our evidence and the measure of our impact comes from real classrooms, real misconceptions, and real learning moments. We don't theorise about impact, we measure it.

From Measurement to Understanding

We understand why a student gets stuck, not just what they got wrong. By identifying students’ misconceptions, we empower teachers and tutors to deliver precise and effective interventions.

Progress Through Collaboration

We partner with leading academics and industry and are committed to sharing what we learn, contributing to peer-reviewed science, and to open-sourcing key parts of our infrastructure for the benefit of the Learning Engineering field.

Equity and Accessibility

We believe the benefits of advanced technology should be accessible to every student and teacher and should positively impact every student equally, in developed and low and middle income countries across the globe.

Safety at Eedi

Eedi is a safety-first organisation

Eedi’s approach to safety is built on three pillars - the pedagogical expertise of our learning designers and human tutors; AToMS; our real-time human in the loop/on the loop (HITL/HOTL) safety and moderation system and our continuous review of data from real live classrooms to pinpoint where more tuning is needed.

Learn more

Credibility from Collaboration

Eedi partners with the education industry’s leading research labs, LLMs and funders

Eedi is the GOAT of sharing data with researchers and enabling new development in AI for education.

[headshot] image of customer (for a bars & pub)
Andrew Lan
University of Massachusetts Amherst

Eedi’s focus on data-driven methods, data collaborations, and large-scale human sourced analytic solutions to pressing educational concerns provides them unique capital to succeed.

[headshot] image of customer (for a financial advisor)
Scott Crossley
Psychology/Data Science Institute, Vanderbilt University

Eedi has been an outstanding research partner and catalyst for advancing the science of teaching and learning. Their commitment to open science is exemplary.

[headshot] image of customer (for a real estate tech)
Rene Kizilcec
Associate Professor, Cornell University; Director, Future of Learning Lab

Our lab is guided by a central goal: AI for human empowerment. Our collaboration with Eedi advances that mission, making learning more personal and accessible for everyone.

[headshot] image of customer (for a construction company)
Mihaela van der Schaar
van der Schaar Lab

Eedi’s open-source data

Eedi has spent over a decade collecting the novel and granular data needed to accurately diagnose and help resolve student misconceptions.

We currently offer our colleagues in the Learning Engineering field seven individual sets of Eedi’s misconception data that can be downloaded and used in their research. Eedi datasets are also regularly used in competitions like this one from August 2026.

Explore and access our data