Multiple independent RCTs show efficacy across several dimensions of teaching and learning
The positive impact of automated feedback for teachers has been validated not just in TeachFX’s internal data and analyses, but also in rigorous studies conducted by independent academic researchers.
Four large-scale randomized controlled trials (RCTs) from leading researchers at Harvard University, Stanford University and the University of Maryland found positive impacts on both teacher practice and student outcomes.
Read more about these studies here.
TeachFX helps develop great teachers
A soon-to-be-published study of teacher candidates at a leading educator prep program found that TeachFX users consistently score dramatically better on their teacher evaluations and report overwhelmingly positive experiences using the tool.
Grounded in decades of education research
-
Students who participate more learn more. Oral participation is a causal mechanism that drives student learning and sense of belonging. (Lotan, 2012; Holthuis, 2012; Michaels, 2008; Bianchini, 1997; Cohen, 1997; Leechor, 1989; Vygotsky, 1978)
And yet, on average, teachers tend to talk 70% to 80% of the time. (Hattie, 2012)
-
Feedback is critical to changing teacher practice (Fallon et al., 2015), but teachers rarely get feedback. The feedback teachers do receive isn’t specific enough to be actionable. Feedback teachers receive is at best subjective and, at worst, is biased and may exacerbate inequities (Kraft et al., 2018, Chalkbeat, 2024).
AI-powered instructional feedback is unbiased, frequent, not time intensive, and proven to change teaching practice (Demszky & Wang, 2023).
-
Most teachers are not satisfied with the PD they receive, citing insufficient time, lack of financial resources to pay for the professional development they want, and learning that is not customized (Gates Foundation, 2014).
-
Only half of teachers report receiving coaching in the past 12 months, with high variance in frequency. (Gates, 2014)
Teachers who have access to regular coaching grow their confidence in their teaching abilities and improve student engagement in their classrooms. (Digital Promise, 2019)
But, coaches work on average with 16 teachers at a time, and close to half of them are also teaching in classrooms themselves. (Digital Promise; Gates Foundation)
-
The students who benefit most from speaking in class — including English learners, students with disabilities, and students from low-income backgrounds — often get the fewest opportunities to speak in class. (Ho, 2005)
TeachFX leads the field in turning classroom audio into actionable insights for educators
Three 2025 peer-reviewed papers authored by TeachFX’s Research & Machine Learning teams highlight our most recent breakthroughs:
Multimodal Diarization (AIED 2025): Our first-of-its-kind multimodal diarization system combines both audio and text to determine who’s talking in class with unprecedented accuracy.
Detecting Opportunities to Respond in Noisy Classroom Transcripts (AIED 2025): TeachFX’s fine-tuned models demonstrate high reliability in detecting high-leverage teaching practices in real-world noisy classroom conditions.
Joint Classification and Quote Extraction of Opportunities to Respond (SEGA 2025): Builds on prior work by introducing a new fine-tuned model that is jointly trained to classify the presence of Opportunities to Respond (OTRs) and extract associated teacher quotes.
