The recorded teacher that talks back. An interactive video platform where instruction is filmed once and a conversational AI runs the tutoring session - over and over, for whoever shows up.
A student in a fourth-grade classroom types an answer into a video lesson. The video answers her - not with a canned "try again," but with the exact explanation her teacher recorded for that exact mistake. She doesn't know an algorithm just routed her there. She thinks the teacher is talking to her. In a sense, the teacher is.
Scene: an ordinary classroom, an ordinary Tuesday, a piece of software pretending to be a conversation.
Who they are nowKyron Learning is an education-technology company with roughly 31 people and one fixation: that the best part of school - a teacher noticing you're confused and saying the right thing next - should not be rationed by how many adults are in the room. The platform takes lessons recorded by real teachers and wraps them in conversational AI. A learner responds by text or voice. The system reads the answer, decides what the student actually misunderstood, and plays the matching teacher response. The result feels less like watching a video and more like being tutored.
It is, the company is quick to say, not a chatbot wearing a teacher costume. The teaching is human. The facilitation is the machine. That distinction is the whole pitch, and Kyron has built a public benefit corporation around defending it.
Built by teachers, for teachers - to help solve the opportunity gap in education.Kyron Learning's founding premise
The uncomfortable research finding behind Kyron is decades old: a student tutored one-on-one tends to dramatically outperform a student in a normal classroom. The catch is arithmetic. There are not enough tutors, there is not enough money, and the families who can buy private tutoring are rarely the ones who most need it. So the benefit accrues to the already-advantaged. The gap widens. Everyone agrees this is a shame and almost no one has the budget to fix it.
Then the timing got worse. By 2022, U.S. fourth- and eighth-grade math scores had fallen to their lowest levels in nearly twenty years, dragged down by the pandemic, while schools faced a teacher shortage at the same moment. More kids needed individual help; fewer adults were available to give it. That is the tension Kyron exists inside - and it has not gone away.
Four numbers that, stacked together, look less like a market and more like a dare.
It turns what has traditionally been a one-way monologue into a vibrant two-way conversation - just like a live tutoring session.How the company describes its interactive video
Rajen Sheth spent seventeen years at Google, leaving as a Vice President of Google Cloud AI - and, before that, as one of the people credited with starting what became Google Workspace. He had spent a career making software that scaled to billions. He left to chase something that, on paper, scales terribly: the influence of a single teacher on a single student. He has said it was exactly that power - one teacher changing one life - that pulled him out the door.
He co-founded Kyron in 2022 with Enis Konuk, founder of Qwiklabs, a hands-on cloud-training company Google had acquired. Between them they had built learning products and AI products at scale. Their bet is contrarian in the way good bets usually are: that scale and human teaching are not opposites, if you let the human do the teaching and the machine do the listening. It is a tidy theory. Whether it survives contact with thirty squirming fourth-graders is the rest of the story.
Teachers record a lesson and, crucially, record responses to the wrong answers they already know students will give. Kyron's natural-language understanding and generative AI interpret what the learner says - reportedly with around 95% accuracy - and serve the right pre-recorded clip. When it can't find a match, it falls back gracefully instead of guessing. The teaching never came from a model. The routing did.
Learners answer out loud or by text and get a teacher's real response to their specific misconception - not a generic hint.
The platform flags in real time what a student got wrong and why, turning every answer into assessment data.
Teacher-built lessons starting in 4th grade, expanding toward 3rd and 5th - the subject where the scores hurt most.
After the Series A, Kyron opened the tools so districts, universities, tutoring firms, and corporate trainers can build their own interactive video.
Four features, one trick repeated honestly: let the teacher teach, let the software keep track.
He left a seventeen-year Google career because of the power one teacher has to change a student's life.On Rajen Sheth's reason for starting Kyron
Rajen Sheth and Enis Konuk start the company as a public benefit corporation, "by teachers, for teachers," with veterans from Google and Amazon.
The 4th-grade math product runs in dozens of schools for 2023-24, including a partnership with Baltimore City Public Schools.
An $850,000 grant funds expansion to 3rd- and 5th-grade math and a study of interactive-video efficacy.
Led by GSV Ventures with Owl Ventures, ECMC, Charter School Growth Fund, Imagine Learning and others. The platform opens to all learning solution providers.
EdWeek Market Brief features Kyron in its K-12 dealmaking roundup as the AI-focused startup raising the round.
A pilot is not a peer-reviewed study, and Kyron's headline figures should be read as the company's own pilot reporting, not independent proof. But the shape of the claim is what matters: when teachers who used it were asked, almost all of them said they'd tell a colleague to try it.
A chart that knows its limits - which, in edtech, is its own small act of honesty.
Free technology in 35+ pilot schools, with a district partner publicly endorsing the classroom fit.Baltimore City Public Schools, 2023-24
Plenty of companies say they care about access. Kyron incorporated as a public benefit corporation, which means the mission - equitable, one-on-one teaching for every student - is a legal obligation, not a marketing line that can be quietly retired when a board gets nervous. The investors backing it are mostly education funds: GSV, Owl Ventures, the Charter School Growth Fund, ECMC's education impact fund. They are betting on the mission and the margin at the same time, which is either admirable or convenient, depending on how cynical your week has been.
Scale and human teaching are not opposites - if you let the human teach and the machine listen.The thesis, in one line
Most AI in education is racing to generate the teaching itself - to have a model explain fractions in its own confident, occasionally wrong words. Kyron took the other fork. Keep the human explanation; automate only the part where you figure out which explanation a particular kid needs right now. If generative tutors win on pure cost, Kyron's teacher-recorded approach will have to justify the extra work. If trust in AI-generated instruction keeps wobbling, Kyron's "the teaching is real, we just routed it" answer could age very well.
Either way, the test is the same one it started with: can a recorded teacher, facilitated by software, actually catch a student at the moment she's confused? That is not a question a funding round answers. It's a question thirty fourth-graders answer, one Tuesday at a time.
Back in that classroom, the student fixes her mistake and moves to the next problem. She still thinks the teacher was talking to her. The company's entire job is to make sure that, in every way that counts, she's right.
Closing scene: same desk, same Tuesday, one fewer student stuck - which was the point all along.