Edtech Insiders
Edtech Insiders
How Abhay Gupta and Frizzle Are Saving Teachers 20 Hours a Week
Abhay Gupta is the Co-Founder and CEO of Frizzle. He was Previously a Product Manager at Coinbase, where he drove $50M of incremental revenue, as well as Tesla, and Meta and helped scale an online e-learning platform to millions in revenue. He holds a Bachelor’s in Computer Science and Economics from Vanderbilt University.
💡 5 Things You’ll Learn in This Episode:
- How Frizzle saves teachers up to 20 hours a week
- Why Agile Learning can transform student progress
- The surprising comeback of pen-and-paper learning
- How AI delivers personalized assignments at scale
- What’s next for Frizzle in subjects and global expansion
✨ Episode Highlights:
[00:00:00] Abhay Gupta on Frizzle’s mission to eliminate grading and lesson planning time.
[00:02:17] From Tesla and Coinbase to EdTech: Abhay’s journey.
[00:04:14] Frizzle instantly grades handwritten math worksheets.
[00:06:56] The story behind the name “Frizzle.”
[00:09:39] Applying product management to classroom workflows.
[00:12:22] Agile Learning: real-time feedback and personalization.
[00:17:46] Why schools are returning to pen and paper.
[00:22:19] Expanding Frizzle to new subjects and countries.
[00:29:26] The future of teaching with AI as a helper.
[00:41:19] How teachers and administrators can try Frizzle.
[00:42:07] The biggest AI trend: schools driving adoption.
😎 Stay updated with Edtech Insiders!
- Follow our Podcast on:
- Sign up for the Edtech Insiders newsletter.
- Follow Edtech Insiders on LinkedIn!
🎉 Presenting Sponsor/s:
Innovation in preK to gray learning is powered by exceptional people. For over 15 years, EdTech companies of all sizes and stages have trusted HireEducation to find the talent that drives impact. When specific skills and experiences are mission-critical, HireEducation is a partner that delivers. Offering permanent, fractional, and executive recruitment, HireEducation knows the go-to-market talent you need. Learn more at HireEdu.com.
Every year, K-12 districts and higher ed institutions spend over half a trillion dollars—but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, then helps you turn them into personalized outreach—fast. Win the deal before it hits the RFP stage. That’s how top edtech teams stay ahead.
As a tech-first company, Tuck Advisors has developed a suite of proprietary tools to serve its clients better. Tuck was the first firm in the world to launch a custom GPT around M&A.
If you haven’t already, try our proprietary M&A Analyzer, which assesses fit between your company and a specific buyer.
To explore this free tool and the rest of our technology, visit tuckadvisors.com.
[00:00:00] Abhay Gupta: If someone from the 18 hundreds time traveled to today, the only thing they would recognize is the school system. And so what Frizzle is kind of pioneering, and that's our ultimate vision, is this concept of agile learning. Where with Frizzle, we can now make the cost of grading zero for teachers, and we can also make the cost of lesson planning zero for teachers as well.
How that works is like the teacher gives a lesson and then the teacher gives that assignment to the student. And so the student can do it on their own time. And once they do do it, they'll basically get instantaneous feedback using Frizzle. And once they get that feedback, it will help them understand where they went wrong.
And then the next step for them would be Frizzle will generate like a personalized lesson plan for them. To address where they kind of fell short so they can practice where they're weak at and really only do the problems that drive the most learning for them.
[00:00:46] Alex Sarlin: Welcome to EdTech Insiders, the top podcast covering the education technology industry from funding rounds to impact to AI developments across early childhood K 12 higher ed and work. You'll find it all here at EdTech Insiders. Remember to subscribe to the pod. Check out our newsletter and also our event calendar.
And to go deeper, check out EdTech Insiders Plus where you can get premium content access to our WhatsApp channel, early access to events and back channel insights from Alex and Ben. Hope you enjoy today's pod
for our interview today. We are here with Abhay Gupta. He's the co-founder and CEO of Frizzle, which does math. Grading using AI, and we'll talk all about it. He was previously a product manager at Coinbase where he drove over $50 million of incremental revenue. He was also at Tesla at Meta, and he has helped scaled an online e-learning platform to millions of dollars in revenue.
He holds a Bachelor's in Computer Science and Economics from Vanderbilt University, Abhay Gupta. Welcome to EdTech Insiders. Thank you Alex, and thank you for having me, and super nice to meet you. It's great to meet you as well. So first off, you know that is a really amazing background and an unusual background for EdTech.
You are coming from Coinbase, from doing cryptocurrencies, from Tesla, from meta, but you also have some EdTech background. Tell us about your journey into EdTech and what brought you into this field and to founding Frizzle.
[00:02:17] Abhay Gupta: Honestly, like, yeah. Normally people with my background tend to just stay in those industries because like you can make money way quicker.
Right. But I personally have a lot of friends who are teachers who actually just recently left the profession because they got burned out and I didn't know this, but like teachers actually work like 50 to 70 hours a week. Right. And living in New York City, right? Everyone talks about like, oh, the investment bankers and the consultants are coming back from the office so late.
But no one talks about the teachers working that hard and like that late, right? So it seemed like a very underserved problem and like underserved market. And so we like talked to a bunch of teachers, like my friends and their friends, and realized that like grading kind of makes up 10 to 20 hours of that a week.
And then we kinda looked at like the existing tools out there in terms of like. There were some essay graders out there. There were some math graders out there that only graded online math, but there was really no tool that could grade math on paper, like handwritten math in classrooms. And so. Me and my co-founder, he was previously doing AI at Microsoft and so we kind of said it seemed like there's a clear opportunity to leverage our tech backgrounds to really help teachers here.
And so that's kind of how Frizzle was born. We started about two months ago in the beginning of July,
[00:03:23] Alex Sarlin: and yeah, here we are so far. Amazing. Well welcome to the ed tech world. So we're really happy to have you. And you know, you, you're talking about bringing an underserved market, which is. Yet, teachers are very overworked.
There's been some really interesting reports about how teachers spend those, you know, 60 plus hours a week that they spend teaching and it's, there's so many different activities on their plate. As you mentioned, grading is a big part of it. There's a lot of paperwork, there's a lot of overhead doing bureaucratic and logistical work.
And then there's obviously lots of actual teaching time and working with the kids. But it is a very, very stressful and burnout friendly profession, which is really not good for any of us. But I wanna ask you about. This solution, you know, you are bringing a tech lens to this, so you're bringing your OCR.
Your tool actually allows teachers to grade handwritten math like worksheets. Tell us how it works. Give us a little walkthrough of what a day in the life of Arizal teacher looks like.
[00:04:14] Abhay Gupta: So say a teacher has like a stack of math worksheets, right? They first have to get those in like an online image format, right?
So what our teachers have told us that they do is they go to like the scanner copier machine in their school and then they just feed in like all the worksheets in bulk and they get it quickly out as a PDF, right? But some other teachers, like, you know, in some of the more underserved areas where they don't have such a machine tell us they use their phone to take pictures of it.
Which they then upload all in bulk as well. Actually, for those teachers, we're building out an app right now where you can take a video of you flipping through the pages really quickly, and from the video it'll be able to take the scans out to really make that like a frictionless possible process for them.
So once they really get all these scans in, then they can literally just go to the frizzle.com, they sign with their Google. And there's like a grading canvas, right? And on the grading canvas, you can upload basically as many worksheets as you want in one go. We've had like teachers upload like hundreds in one go, and you just like drag and drop all the files on there.
And then in about like a minute or so, you'll get every worksheet graded with like red ink feedback on the page. Just like how a teacher would mark it, right? So not only is this helping teachers, it's also helping students because they're getting this like individualized person precise feedback that they normally just can't get because the teachers don't have the time to give it back to them.
And so then once the teacher. Seize this feedback on the page and they're good with it. They can then like just export it out as like a PDF and then like either email it back to their students for like, some of the older students or print it out for some of the younger students and give back like a hard copy.
[00:05:36] Alex Sarlin: So it's, it is really a play to sort of very quickly digitize and then be able to give very rapid feedback and decrease that turnaround time. So the, so teachers have less time grading, but also students have much less time before getting their feedback. From their work.
[00:05:52] Abhay Gupta: Exactly. And so it gives, it shortens that like, you know, learning feedback loop to where students get that feedback instantaneously, essentially, and then they can really reflect on the feedback and understand before getting onto the next lesson, which usually kind of builds upon the past lesson anyways.
[00:06:06] Alex Sarlin: And because in many math worksheets, the students are showing their work. They're not necessarily just writing the answer. They're showing steps, they're showing breakdowns. There's a lot of opportunity for that very detailed feedback. Exactly.
[00:06:17] Abhay Gupta: And so Friso can actually call out exactly in the student's work where they went wrong.
So for example, right, if you're solving like a equation and you solve for XXI three, and then you substitute X back into like a different equation. But the X that you saw for was actually inaccurate. Friz will actually call out that you messed up at that X step rather than every other step before that.
So we can really actually call out that propagation of errors too, and really help pinpoint where the student went wrong.
[00:06:40] Alex Sarlin: That's exciting and I have to ask you, you're called Frizzle and you know, we've talked on the show about how Miss Frizzle from Magic Schoolhouse is sort of a really amazing avatar of a certain type of innovative education.
In that case, it's immersive education. Is Frizzle named after Miss Frizzle?
[00:06:56] Abhay Gupta: It is, and actually like in my own elementary school, we would watch Magic School was all the time and it really is like an ode to Miss Frizzle and you know, bringing that kind of. Energy and that playfulness back to the classroom.
[00:07:07] Alex Sarlin: Yeah. The Playfulness Magic School Bus is all about playfulness and sort of immersion. I think it's a really great, it's a great name. So how have people reacted? You're using and just to fill in the gaps, if people are trying to envision what this looks like, the worksheet becomes an actual, an image, and then teacher literally has written feedback all over the image.
It looks like it's been written on by a teacher, but it's actually the teacher is using AI to put their own feedback in. Organize it, get it in exactly the right place it, it basically, the resulting output looks a lot like a graded paper that we are all used to. How do people react to, how have your, your audiences, both teachers and students reacted to Frizzle so far?
[00:07:44] Abhay Gupta: I mean, honestly, we've went around the country over the past two months really going to conferences to meet teachers, you know, random states that I wouldn't have said I would've gone to before, like Idaho, Montana, Oregon. But it's been amazing to meet teachers and really understand their workflows. The overwhelming response has been, where were you 20 years ago?
Like that's kind of what we've heard from all of them. And so it's like, I mean, we told them it's like we're here now. So you know, that's kind of what we're trying to help out from, but it's been overwhelmingly positive. I think there have been some cases of feedback where we've actually used that to iterate ourselves.
So like some teachers have told us that like. They like to grade themselves. 'cause it helps them understand how their class is doing. And so we wanna help teachers also be better teachers. And so what we're actually building out right now for teachers is like, once we grade your classes work, we'll tell you, okay, the class is doing really well on these topics, but they're kind of struggling with these ones.
And we'll even break it down student by student, right? Like John and Jill, they're great with exponents, but they're kind of struggling with like division. So maybe you should like. Focus on that with them. And so really giving teachers these data-driven insights so they can be better teachers to really help kids learn better.
[00:08:49] Alex Sarlin: Yeah, and I mean, to use a little math metaphor there, it's sort of like you're giving the derivative, right? You're looking at all the the all the students' work, and then pulling out an insight that's sort of a higher level insight and saying, oh, we looked across all of your students and we found this common misconception, or this group of kids are all dealing with this problem in this way.
You might wanna pull them out and do a small group that is incredibly powerful. In a way that's often what teachers are truly trying to get by grading all those, you know, 30 papers by hand. They're trying to figure out what each individual kid is missing, but also what the class is missing and what each group is missing.
And AI is very good at synthesizing data like that.
[00:09:24] Abhay Gupta: Exactly. And yeah, like normally a teacher, like, you know, with 15 kids can really start to recognize the patterns in their head. But like when there's like a hundred kids in like five sections, like those patterns just get lost in, you know, all the noise.
So we really wanna make that very clear for them.
[00:09:39] Alex Sarlin: One thing that strikes me as I hear the way you're doing this, you say you're going around the country, you're looking at workflows, you're likely are interviewing teachers and sort of seeing exactly how they do things now and what you're replacing. You come from a product management background at some very big tech companies and I, I'm curious, you know, I bet you bring.
Certain lens to this work, just with that product background, that with that specific tech product background that you maybe others in ed tech don't always think about. You've mentioned the concept of sort of agile learning, like we have agile product management, and you have this concept of agile learning.
That's really interesting. Can you explain that to us?
[00:10:11] Abhay Gupta: So in software development, how software used to get developed in like, you know, 20 years ago was this like waterfall method where you would basically get like a product spec or like a requirement that you would give to the engineers and they would just build it out and you really wouldn't talk and like iterate in the meantime.
It would just kinda be like a straight linear process. Right. And then like, you know, about 10 to 20 years ago we switched over to this process called Agile, where it's like, it's like a circular thing where engineers get these requirements. They're making like progress week over week, but they're also interacting with like the product manager helps, you know, make sure what they're building actually helps the user.
They're also working with designers to make sure, you know, all the constraints are really properly balanced. And so that kind of helps teams nowadays shift software faster than ever and more efficiently than ever. And so. Right now, my whole thing is like in the school system, right? If someone from the 18 hundreds time traveled to today, the only thing they would recognize is the school system, right?
Because it's the exact same as ever before. And if you look at like the school system, it follows a waterfall learning process where it's like the teacher gives a lesson in the classroom. The student gets a homework assignment to do, which they do in about a week, and then the student gets that feedback in like that grade.
If they do get that feedback back in about like one to two weeks after that, and by the time they really reflect on that feedback and understand what to improve for next time, the class is already on a unit like ahead. Right. And so like. The student is kind of like lost in the loop at this point, right?
And so what Frizzle is kind of pioneering, and that's our ultimate vision, is this concept of agile learning. Where with Frizzle, we can now make the cost of grading zero for teachers, and we can also make the cost of lesson planning zero for teachers as well. And so now how that works is like the teacher gives a lesson and then the teacher gives that assignment to the student.
And so the student can do it on their own time. And once they do do it, they'll basically get instantaneous feedback using Frizzle. And once they get that feedback. It will help them understand where they went wrong. And then the next step for them would be Frizzle will generate like a personalized lesson plan for them to address where they kind of fell short so they can practice where they're weak at and really only do the problems that drive the most learning for them.
Right. And so it's kind of this like learning feedback loop and. The teacher will be able to guide each student on their own personalized learning journey on pen and paper thanks to Frizzle, and that's kind of our ultimate vision for Frizzle.
[00:12:22] Alex Sarlin: Yeah, I think it's a really intriguing metaphor. Partially because, you know, some of the benefits of agile development in in product management and software development are that you catch errors early, right?
Where you. Even product market fit issues early. Like if there's something you're doing wrong, you don't only find out when you're done with the project. That's sort of the the, and I think that's a really good metaphor for especially math learning, which tends to be so cumulative, right? If you get to the test at the end of a unit and find out, so many of your kids haven't understood some major concept, you really as a teacher, you really wish you had known that much, much sooner.
And I think it's a really good parallel to how agile. Tries to improve on waterfall. It's like if you're meeting constantly, if you're always connecting in a very frequent loops, like you mentioned, you can catch errors way, way earlier and fix them in real time rather than only finding out about them too late.
And that seems, I've never heard anybody bring that metaphor to education, but it makes sense to me.
[00:13:15] Abhay Gupta: Thank you.
[00:13:15] Alex Sarlin: Yeah.
[00:13:16] Abhay Gupta: That's why we're
[00:13:16] Alex Sarlin: pioneering it,
[00:13:17] Abhay Gupta: right?
[00:13:17] Alex Sarlin: To
[00:13:17] Abhay Gupta: bring
[00:13:18] Alex Sarlin: it to education. Yeah. So how are you thinking about measuring success at Fri All? I know you're only a couple months in, and this is a little bit of a a, a forward thinking question, but you mentioned on the site that one of the core goals is saving teacher time, right?
That's how you say. You said teachers are spending 10 to 20% of their very long work weeks doing all this grading, especially math teachers, and you say you can save them time and give them more time. For the classroom, for lesson planning or for, you know, for teaching, for connecting with their kids. Are there other metrics of success?
Another product management question, right? Are there other metrics that you are trying to hold as sort of your north star, your goal as you progress?
[00:13:51] Abhay Gupta: Alongside teachers. We've also talked to a bunch of principals and superintendents, right, to understand like what are their priorities as well. And like what they've told us is like right now, the teacher shortage is as bad as it's ever been, right?
One in eight teacher roles is vacant right now, right? So teacher time is also the most valuable resource for these school administrators, right? Like I remember like a director of technology for a district was telling us that he had to do like a cybersecurity training for his teachers, and he only got 30 minutes for the entire year to really teach teachers about cybersecurity, which was not enough.
Right. And it's because their time is so valuable. And so really unlocking that time is our number one metric for both teachers personally and then also for school administrators as a whole. But I think some other tangential metrics that we were looking at as well is like. Teacher burnout. Right? Teacher turnover.
Teacher retention. Right. Schools are dealing with these problems as well because a lot of them, like my own friends, have left to go to do like recruiting or sales or something else where they're treated better, right? And so we do really wanna prevent that outflow of teachers from the education space into other industries and actually bring back teachers to the education space by making a better.
Life for them.
[00:14:55] Alex Sarlin: Yeah. And then I imagine, you know, when you're doing something that is dedicated towards teachers, administrators, sort of school systems, reducing teacher burnout, there has to be a sort of paired metric of, but the grading still has to be at a level in which the students are learning, you know, as well as they would with, if the teacher is spending, you know, 10 extra hours grading the papers.
I'm curious how you do that. Like how do you think about, because obviously, you know, in the, in the ed tech world, in the AI world in general, people are trying to figure out what things AI can do best, what things humans can do best, and there's always gonna be this sort of push and pull that's just a lot of speculation about what the jobs of the future are gonna look like in that particular vein.
And I think doing automated grading of math worksheets, in some ways that feels like a. Pretty clear idea of the type of work that could be partially or almost entirely automated at the same time. There is definitely value add that teachers are doing when they spot certain things that a kid is doing wrong, when they bring their expertise, when they bring their experience.
How would you think about delivering feedback to the students in a way that really maintains the quality that the teachers would?
[00:15:57] Abhay Gupta: Yeah, I mean, so how we've thought about this is like at the end of the day, each teacher kind of really has their own grading style and feedback style. And so we wanna meet teachers where they're at, right?
So like some teachers like to like walk through their students through their solution and like problem. And some teachers like to really prompt their students with like questions rather than just tell 'em the answer. Right? And so on frizzle, we're actually creating like this custom grading cell option where teachers can specify how they would like to grade.
And based off of that it will grade. So like. Say a teacher wants to walk through the solution, then the friz will give that feedback for each question on the page. Like, okay, hey John, like, here's where you messed up, and like, this is how you should have done it. Versus like, if a teacher wanted to be more like, you know, question based, uh, feedback, then Frizzle could be like, okay, hey John, did you look at this step?
I think you might've missed something here. So we really wanna meet teachers where they're at and just support them in their existing style. Like, I think for us, we've talked to like so many teachers. For them where current tools fall short is like they're not built for teachers. So we're really taking their feedback to heart and building from the ground up for them.
'cause they are the people we're trying to help at the end of the day.
[00:17:00] Alex Sarlin: Yeah, and using that metaphor of agile learning, it makes me wonder if there may be some kind of loop to test what type of feedback might actually be more beneficial to the students, right? If the student is able to get feedback much more quickly because of a tool like frizzle, and then redo the homework or try a different question with a similar concept, you potentially can actually be able to give data-based feedback on what.
Type of responses that the teacher is giving that, that are actually making a difference, which is something we've never had access to before. It's always been such a black box right in, in that, in the waterfall type system you talk about where student hands in their homework and they get it back, you know, a week or two later with the red lines on it.
You don't know. It's been very hard to know whether those red lines are even red and if they are red, whether they make any difference. But potentially with AI we could start to know that.
[00:17:46] Abhay Gupta: Exactly. And like the big unlock for us as well is like, we wanna support it on pen and paper, right? Because it's like.
We've heard from like hundreds and teachers in schools, like they're all moving back to pen and paper. And it's because, you know, during COVID, all these schools across the nation adopted like, you know, one-on-one Chromebooks, one-on-one iPads for their kids in the classroom, and now five years later, they're looking at the student outcomes.
And the student outcomes are terrible, right? Like the students don't retain anything. They're like, what happened? Right? And so now all this research is coming out that is proving that students learn way better on pen and paper and retain way better on pen and paper, right? Like that physical motor ability really helps them lock that in their brain.
And then additionally, like. When you have like a device in your classroom, like students are just distracted, like they're watching YouTube, they're playing games, right? The, these kids are young, and so they don't have that level of like prefrontal cortex development to where they have that willpower to not engage with these devices.
And so schools are just doing device bands. So like now it's just like the complete opposite. And then finally, like teachers have told us that. It's way harder for a student to like copy and paste something in the chat GPT when they have to like hand write it versus like, you know, piping something up.
And so even as a way to prevent cheating, they're doing this as well. And we see this at the collegiate level too. And so it's a clear trend and we wanna support this trend. And I mean, I myself also learned on pen and paper and I think it was like, you know, an effective, I like I remember so much of my education still to this day.
And so I do wanna really support that for the kids of in the future as well.
[00:19:11] Alex Sarlin: That's such an interesting insight. You know, it strikes me as I hear you talk about some of the benefits of pen and paper, both the integrity benefits that you could, it's something you can, you can sort of lock down, but also some of the learning benefits there is research about you, you know, the tactile actually working on paper can engrave things in your mind in a different way.
There is definitely some, some research about that. You know, it, it brings up a really interesting sort of tension in Ed. In general, right? Is that, you know, we're trying to bring technology solutions into classrooms to enhance learning outcomes and, and efficiency. And there are other metrics as well, but we want better learning.
Yet sometimes technology itself just swims directly against the current tech. Just the presence of technology can be distracting, can be disorienting, can, can take people away from certain types of learning outcomes. So what you're doing at Rizal, and I think, you know, it's, it's potentially an interesting trend for the AI era is.
Allowing the sort of analog and digital worlds to flow very seamlessly back and forth to one another.
[00:20:05] Abhay Gupta: And that's what our goal was really to like support teachers in their natural workflows, right? We don't want 'em to switch their behavior, right? Because students do learn better on paper. And that's just like even me, like when I have to write a to-do list, it's just way more effective when I write in my journal versus I type it out on my like notes app, right?
And so it's just something that's like a human element to just like being human. And so I think we wanna preserve that for teachers and for students.
[00:20:29] Alex Sarlin: Yeah, it's really interesting. And then I'm curious, you know, a lot of what we see on the Frizzle site and sort of your, your use cases are, are often very standard looking worksheets.
'cause again, you're asking people not to change their workflows. This is a very commonly used approach. But I'm curious if there are other, you know, paper-based types of assignments that you either are already seeing people use Frizzle for or something that you might. Anticipate them using for the future?
I mean, do you see people using it for drawings or people doing schematics or diagrams or any other sort of paper-based assignments that this would be like a perfect match for?
[00:21:01] Abhay Gupta: So one interesting tidbit is like we had some teachers who were elementary school teachers upload some worksheets where it's like.
There's a lot of like images and drawings on the page, right? To make it very kid friendly for some of these younger kids. And so we had to really train our AI to like, you know, those images are not part of the problem, you know, like this is where the problem is. So like being able to understand those, like more kid friendly worksheets was an issue for us at first, but then we were able to train that AI to do that effectively.
And then I think another thing was also important was we actually were working with a teacher from the German school in Brooklyn. So they do their math in German, and actually they uploaded German worksheets to our Frizzle tool, and Frizzle was actually able to handle it. And so now we do actually support all these different languages as well.
So with AI, we're really unlocking like a new level of intelligence and support here that can really transcend globally. I did actually wanna call out that we're, we're working with the school in Namibia, one of the leading private schools in Namibia. Where they use a marking scheme. 'cause you know, there they call it marking instead of grading from the UK called the I-C-G-S-E marking scheme and it's like the UK Standard Grading scheme.
And we built out like a custom I-C-G-S-E grading engine for them based on like, you know, it's like their version of AP essentially and like to basically work, work with their formats. And now we're using that to really expand in like, you know, South Africa and Namibia and also like the UK and Europe as well.
[00:22:19] Alex Sarlin: Yeah, so I, what I'm hearing you say is that there's different ways in terms of both the grading schemes and use cases for sort of paper-based technology. It's a potentially an expanding set of use cases. Like you started with very, very standard math worksheets. You already have begun to, you're two months in, already have begun to adapt to your.
Worksheets with lots of fun images and cutesy things for for kids and make sure that the AI knows how to handle them. You've already had to adapt to multiple languages and multiple great marking schemes. And I envision in the future as just with more partners, with more use cases, there's lots of different.
Potentials because AI is such a flexible technology that as you start to have new use cases, it's not like, oh, we built this huge system and we're locked in. It's only, we can only do this one thing. AI is so flexible. This something I'm really excited about in the ed tech world in general right now. And I can imagine easily, uh, other, other use cases that would come.
I mean, I can imagine this being used in art class, right? Another very paper-based, in many cases, very paper-based. Learning method where people are drawing and it can give incredibly specific feedback about, about the drawings. Uh, that would make total sense and that could be a really exciting product in and of itself.
[00:23:26] Abhay Gupta: Yeah. And then also I did wanna call out, like we actually are actively expanding to other subjects, um, because you know, like we've had science teachers come up to us and be like, can you guys build this for science too? That is very top of mind. Like the reason we chose to start with math is because math is objective, right?
And so it's like there is a right or wrong answer. And so we really wanted to establish that trust with teachers to prove to them like AI can grade before going into like more some of the subjectives subjects as well.
[00:23:53] Alex Sarlin: Gotcha. Yeah, that makes sense. So if you sort of push this out and say, okay, we're gonna go into mo, we're gonna go from math to science and from science to English language arts and, and to other subjects.
Do you anticipate a world, you know, if, if Frizzle and other ed tech tools like it continue to grow, do you anticipate a world where. The role of a teacher really changes very much away from this sort of hardcore grading. Looking at every student's work very carefully to much more of a sort of orchestrator, higher level.
Looking at how all the pieces fit together rather than diving deep into any individual assignment. Is that sort of the future that you feel like tools like frizz are driving us towards?
[00:24:29] Abhay Gupta: Definitely. I mean, so like there is this fear with ai, right? Like is it going to replace my job? Right? And we've had that fear from many teachers across the country and like what we've told them is like.
At least for teaching. There's nothing to be worried about with AI because, you know, at the end of the day, teaching is like a human to human connection. It's a human to human relationship, right? We learn from our elders. That's just like a tale as old as time from the beginning of like, you know, caveman days, right?
So teachers will always be there. But what we envision AI kind of becoming is like a helper teacher, right? So that it becomes very easy to manage students and help students on a one-on-one basis at scale, thanks to ai, right? Taking care of all the grading, all the lesson planning, the personalization, the grouping, kids by skill level, all of that will be automated way to where like the teacher can really just go around in class and work with students on like a one-on-one basis and not really have to worry about anything else.
And, and our students will get that, like individualized tutoring and learning that most wealthy parents pay like thousands of dollars a year for right. And we wanna really democratize that for everyone.
[00:25:29] Alex Sarlin: Yeah. And I guess a, a related question would be arguably one of the reasons why we've gotten to the sort of current system of, let's say, math education in the first place, where every student has the same assignment every week, which is, is pretty standard.
It's not, not not always true. You know, teachers always want to differentiate. They want to be able to adapt to different students' experience level their interests, but. There's only so much time. Teachers are already working, you know, 60 plus 70 hour work weeks. And so you end up with, you know, every student gets homework one A with their, the same six questions and the same extra credit question at the end.
Do you anticipate that if grading becomes. Commoditized or becomes very quick to do, will teachers start to be able to differentiate homework? Would it be that like these students aren't even getting the same assignments as everyone else in the room because they can actually get assignments that are differentiated and the grading can be done just as quickly the teacher doesn't get any benefit having to check the same, the same question over and over again because they're not doing the grading in the first place.
[00:26:28] Abhay Gupta: No, exactly. That is exactly what will happen because it's like, so even me as a student, right? Like I remember in school I would be given like a problem set, right? I already knew how to do like half the, the problems, right? And so I'm like, why am I already having to do these when I already know how to do this?
Right? And so it's like, and you know, like kids always complain about having too much homework and all that stuff, right? And so it's like you really only want our kids to do what they really need to do, right? Like if they already know like exponents, like why do they have to do more exponent work, right?
They should be learning the next topic that they. Struggles with. So it is really about personalizing that assignment for each student to like, if John is great at exponents, but he is struggling with division, let's give him a division assignment. And if Joe is struggling with exponents, but he is great with division, we'll give him an exponent assignment.
Right? So that will be possible now with ai, with Frizzle, right? And like that is kind of the future where a teacher, especially when they have like, you know, 30 kids plus, they can't keep track of that all in their head and they just don't have the time to even do that themselves, right? Being able to do that at scale will really help meet students where they're at and get them to the next level in the minimum amount of work required to do that as well.
[00:27:32] Alex Sarlin: Yeah. It's also interesting to combine that, that idea of, of personalization or adaptivity differentiation with paper because I think that's something that there, there are ed tech tools, some of which have actually been around quite a while, that try to do some. Elements of exactly what you're saying in terms of giving students assignments that don't, that try to skip prerequisites or try to meet them at their level, or that in of itself is not entirely new to EdTech.
What I think is new is the idea that that would be done on paper, that every student would be getting a different paper assignment and that that different paper assignment. Could be graded and, and iterated on through a digital app. It's, there's something very interesting about, as you say, working within teachers' existing workflows, but then being able to enhance and speed up their processes so that all these things become possible.
[00:28:17] Abhay Gupta: Yeah. I mean, me and my co-founder, we always joke, we're like, we're digitizing paper for teachers. Um.
[00:28:22] Alex Sarlin: Yeah, that's kind of our joke too, but it's not a crazy idea. I think you have a product background and it feels like one of the big insights that you're coming to with this is that even in this, there's already a a, an era of there, there's a little bit of a tech backlash, maybe more than a little bit of a tech backlash right now against cell phones in schools, against, against education technology, in some cases against digital platforms.
But even before that, there was a feeling of, well, teachers have set up all of these complex existing workflows around. Paper assignments and they may be going back to that or even doubling down on that. There's actually a great New York Times editorial from Clay Shirkey as the Vice Provost at NYU, talking about how at the university level we are gonna start to see a huge rise in, in the sort of Blue Books, oral exams, you know, written exams, handwritten exams, like things that are really, really.
He said basically the professor class has tried to adapt their assignments, but they haven't really succeeded. And the best thing they're thinking to do is really go back to a much more analog world, but you could still enhance it with digital. And I think that's one of some of the interesting story that you're bringing to it with Frizzle.
[00:29:26] Abhay Gupta: No, exactly. And we've heard the exact same thing from professors as well too. And it's like, we know that's coming, but it's like with that right, there will be a huge stacks of paper that need to get graded, that need to be worked with. And so we're really trying to get ahead of that curve. If professors or teachers are hesitant to go back to paper because they know it's gonna be so much work, we don't want the amount of work it'll be to be a blocker for them to be able to support their students on pen and paper.
Right? And so that's why we really wanna unlock that access for them. And I do wanna call out that like the administration, the White House administration recently released like an executive order, right? On how to like effectively implement AI in classrooms. We're actually partnered with the White House to make that happen.
Now, I'm actually attending the White House next week to join like an education round table to speak on this, but they know that the existing tools out there really just don't meet teachers where they're at. And so the reason we were able to partner with them is because they've kind of looked at Frizzle and understand that.
We're just like, you know, slotting into exactly where teachers are working and like how it's like going. And so I think that's really helped us stand out among the sea of noise as well.
[00:30:29] Alex Sarlin: Yeah. I have a slightly provocative question, but I'm curious about your answer to it. You know, I mean, I think one thing we've seen at the, at the higher ed level, especially.
Over the last six months or so is this feeling of everybody is a little confused about what is allowed with AI and what is not allowed with ai. Policies are not always clear at the university level. In K 12 schools. There's not always clear at the, at the site or district or state level, there's just a sort of a lot of confusion.
And one of the things that's started to come up is this feeling of, well, who is allowed to use ai? And one thing that that strikes me as sort of interesting about the potential future with Frizzle is you may get to a world where students are being forced to do their work on paper. And then the teachers and professors are grading it with ai.
That's sort of the vision. Do you think there's gonna be a little resentment that is fermented there from the students who say, why should we have to write on paper if our teachers are not even writing on paper, they're digitizing their written responses to us, but we have to do all the writing? How would you address that?
Because I've seen some of that complaint already starting to happen.
[00:31:25] Abhay Gupta: No. Yeah, totally. And that is something we've heard as an issue and like the way we've kind of talked to 'em about is like. So if you look at the current state, right? Students aren't getting feedback for like majority of their assignments, right?
They really only get feedback for tests and being able to, as a student and as a parent, like if your child is getting feedback for every single question for homework, right? Like wealthy parents pay thousands of dollars a year for that level of personalized training, right? And so if the teacher can now bring that to every student in every like.
Neighborhood, like there's really nothing to complain about because it's really only helping the kids and the parents, right? Because the kids are more engaged, they're learning better, and the parents don't have to pay out of pocket for like out of school, like learning centers and such to really make that a reality as well.
[00:32:08] Alex Sarlin: I think there's some truth to what you're saying, but it, it doesn't necessarily address the resentment a student potentially might feel in saying like, I think the parents, to your point, the A parent would say, wow, you're getting way better feedback and I maybe I can even see the feedback you're getting on your work.
This is way better system. But sometimes the incentives in education are not always perfectly aligned and I think some students might find it easier or more. Fun or engaging to do things not on paper because they have access to all these tools and they're sort of being pushed out of that. There's no right answer to this.
There's no good answer to this right now. I think it's just the world we live in right now, but it's really interesting tension that I think is happening as AI enters the, the education system. This feeling of like, what is the point of school? I. Some interesting articles recently about how like, students are complaining that they feel like, well, if teachers are banning all of these tech tools or AI applications, or in, in your case, maybe even calculators, right?
If you're doing everything on paper, you don't have access to Google or calculators. The teachers are saying yes, right? That's how you learn. That's how learning works, right? The students are saying, you seem like you're taking away all our tools. You're taking away all our, all our devices, you're taking away all the things we know we're gonna have as adults.
What are you trying to do to us? I'm curious how you, how you sort of like navigate that.
[00:33:21] Abhay Gupta: Yeah, yeah. Totally. So also on the last point where it's like, how will students feel like we have thought about this, where it's like, well, we wanna enable now with ai, right? Say a teacher gives an assignment, right?
And the teacher gets that feedback back that like, okay, John is really struggling with exponents, right? And the teacher knows that John loves Justin Bieber, right? With AI we can create like a 32nd TikTok where Justin Bieber is teaching John about exponents, right? So we can make it very engaging for these students as well.
It doesn't have to be all pen and paper, right? In terms of the work is just pen on paper. But the way they learn can be interactive and engaging and really uniquely personalized to each student as well. To make it more engaging for that.
[00:34:01] Alex Sarlin: I like that distinction. It makes a lot of sense. You're distinguishing between the working and the learning and maybe the homework or the assignments that are being done in in worksheets would be sort of occupy a different space than the, oh, I actually need to bone up on this and learn this concept.
That makes a lot of sense. That. It's interesting. You know, we, we try to chronicle all these different use cases in AI and education and it's, it's a very complicated ecosystem because it can affect almost any aspect of education and learning. Uh, formal or informal, any subject. Like there's almost nothing in education that couldn't be changed through the use of ai.
So putting together the pieces and saying what happens where is complicated, but I think that your, your point is well made, which is that there's a difference between. The teacher wanting the student to show their work on the homework and write out all the steps and make sure that everybody knows that they can do it versus.
If they're confused about a concept that nobody should care where they get information about that concept, they can use any AI tool. But at the end of the day, if they can't come back and app apply it on that worksheet, then there's an issue.
[00:35:00] Abhay Gupta: Exactly. And yeah, in terms of like how we envision future of learning to be right with ai, it's like at the end of the day, learning is kind of a way to build like logic and reasoning skills for kids to be able to enter the workforce to just be able to solve any problem that comes their way.
Right? And so. At its core, like at the beginning, it's important that a kid doesn't have access to like AI and calculators and tools like that. So they can build that critical thinking skills themselves. But once they have those skills developed, then they should have access to those tools so they can like leverage them and, you know.
Solve the way bigger problems using those skills and the tools in combination. So it's like we really wanna establish that foundation for kids to develop those skills themselves without tools, and then be able to leverage those tools in the future and pair them with those skills to solve way bigger problems.
[00:35:47] Alex Sarlin: So you're entering the ed tech world from really interesting big tech companies at a very, very interesting moment. And you've mentioned that you're going to the White House already. You, you've done Y Combinator, you're sort of in a moment where all eyes, I, at least from my experience, you know, a lot of eyes are on education, a lot of eyes are on AI and education.
I'm curious what that experience has been like, just like writ large, going from working at a meta or a Coinbase to teachers in Montana. That's like a, a pretty big change in, in your sort of worldview. Tell us about what that's been like and sort of what it feels like entering the EdTech ecosystem at this moment in late 2025, sort of as the, as the a AI world continues to boom and confuse.
[00:36:30] Abhay Gupta: Yeah, I mean, so like, you know, I've been an avid student myself growing up. I loved my teachers. I was always a kid that would always, you know, ask questions and you know, get the homework done quickly. I also went to like a boarding math in high school. So, you know, I love learning and I think the reason I became a product manager is because.
You talk to your customers and you learn from them about their problems and how you can help like solve them, right? And so joining the education space has been really wonderful because I get to go around and talk to these teachers one-on-one and just learn about the workflows. Like, you know, having been to Montana, meeting teachers there, just learning about the struggles they're facing in the teaching space, it's just like way more fulfilling and way more impactful.
And that's kinda what keeps us going as well. We kind of did the math and like with Frizzle, as we expand to more subjects, we really have an opportunity to save teachers like 4 billion hours a year across all subjects just alone in the us right? Nothing has been helpful for teachers, not even like the invention of the computer.
And we really have like a once in a lifetime opportunity here on our hands, and that's kind of what gets that motivation for us every morning and like keeps us pumped to work on this every day in and out and like be excited to meet these teachers and like people in education are wonderful, right? They bring very positive energy.
Everyone comes from the same good place of just like, you know, wanting to help kids and wanting to help society. And so I think it's been like an amazing experience so far.
[00:37:49] Alex Sarlin: Yeah, that's really wonderful. I mean, it's, it's a system that has needed a lot of optimization and has needed a lot of of care. Uh, you know, I think that there aren't probably.
That many product people who have gone into classrooms and followed a teacher through their grading workflow and said, what do you do when this happens? What do you do when that happens? I think it's a part of our social fabric that is so important and yet not enough attention is paid to it. So I, I think it's really interesting that you're finding it so gratifying and yes, people in education and EdTech are amazing and frankly, I bet once in a while you have overlaps between those teachers in Montana and the, and the crypto traders or Tesla owners that.
That were your, your previous clients. I'm sure some of those are the same people.
[00:38:29] Abhay Gupta: Definitely. No, definitely. And even their like loved ones are like teachers or something, so we always, you know, end up being someone.
[00:38:35] Alex Sarlin: Exactly. So what's next for Frizzle? You know what? You talked about expanding subject, you talked about expanding languages.
We talked about expanding into, into different types of grading schema and going international, which is a lot for our company. Again, only a few months old, you know. What do you anticipate, I guess one question specifically is, we've talked a lot on this podcast about how in the AI era micro companies are in, you can get a lot done with very few people, and I'm curious how you're thinking about that as a, as a new startup and a relatively small startup, quite small startup, do you anticipate.
Wanting to, you know, as you get more customers, as you get more business, do you need to grow or do you, are you gonna try to stay very lean and sort of use AI to enhance what you can do?
[00:39:15] Abhay Gupta: Yeah, I mean, so I will say this now, like my co-founder, who's the CTO, 90% of the Cody writes is with the help of ai. And he's able to like write so much more code now because of that.
Right? And so like AI is a tool that we are using really to like leverage us at scale. I think we will end up growing though a bit because we do wanna get this out there in the hands of teachers as quickly as possible, right? So that they're ready for this switch from back to, you know, electronics to pen and paper.
Not delay that switch either for the sake of our students. And so we are focusing on growing like on the engineering side. And then also like the, the sales side. We wanna also include in and hire like former teachers, former school administrators, really, you know, the ones that have know the system in and out, and can help us really help spread the mission as itself.
Like at the end of the day, like our number one priority is hiring people who are passionate about the mission. Helping teachers globally, and that's kind of what we're looking for really. But yeah, next, next for us is just building that analytics software for teachers, right, to understand where they can help their students on an individual basis.
And then also adding in those other subjects and then also including in that. International expansion with the ultimate goal of, you know, agile learning and bringing that to every school.
[00:40:27] Alex Sarlin: Yeah. Then that video feature where you can video sort of flipping through a whole bunch of worksheets and it can capture all of them.
That is, that is a really neat, that is a really neat idea. Well, luckily, I mean, luckily on, luckily, this is sort of a funny subject, but the flip side of teacher burnout being so high is that there are often some incredible educators who are looking to move into ed tech, and if you're looking to hire teachers to do.
Amazing work for your team. There are so many incredible teachers that are, I, I, you know, we have met over my journeys, but I think anybody in EdTech has seen these unbelievably talented, devoted, really professional people coming out of the classroom saying, I still care about kids. I still wanna change the education system, make it better, and make every kid have a better learning outcome.
And I'm looking for, you know, ways to do that at scale. So I don't think you're gonna have any trouble finding some amazing teachers for your team as you expand.
[00:41:19] Abhay Gupta: I definitely appreciate that and if there are any teachers out there, especially math teachers, at least at the moment, would love for you to try out frizzle and see if it can help save you 10 to 20 hours a week as well.
If you just go to frizzle.com and if there's any school administrators, like principals or superintendents and you think this would be a great fit for your teachers, I'd definitely love to chat as well. My email is a bay@frizzle.com, so feel free to reach out and we can definitely chat more.
[00:41:43] Alex Sarlin: Fantastic.
So we, we, we tend to end our interviews with two questions, and I'm curious about your answers to this, especially because you are somebody moving into this ed tech space at this particular moment. So what is the most exciting trend that you've seen in your journeys, in classrooms, in your talk with investors or with other founders?
What's the most exciting trend that you see in the ed tech landscape right now then you think our listeners should keep an eye on what's coming right around the next bend?
[00:42:07] Abhay Gupta: Yeah, I mean, so I think the big thing is like. Now school administrators and like especially at their level, are thinking about how do we leverage AI in our classroom, right?
So the script has really flipped where instead of ed tech companies going to schools, it's more like now schools are looking for AI solutions themselves and like what to adopt. But there's a lot of noise out there and there's not a lot of good guidance out there. And so I think it'll be increasingly important for tools that leverage AI in the education system to make it clear to educators and like have that, you know.
Educating the educators really on how to leverage AI effectively. And so I think educating the educators will be very important over these next few years on like how to effectively teach classrooms with ai.
[00:42:50] Alex Sarlin: Yeah. Oh, well, you, we, we saw Open AI team up with the American Federation of Teachers and the United Federa Federation of Teachers to create a giant AI training center in New York, just opening now.
And I think that's one of many such initiatives to start to raise the awareness and knowledge and ability to decide on what tools to use for, for everyone. And that's a really interesting one. And what is a resource you'd recommend? So, you know, I, this could be from your product management background, it could be from your education transition.
What's a resource you would recommend that you find very useful for learning about any of the topics we discussed today? Whether that's grading, whether that's teacher burnout, anything. What would you recommend for our listeners to look at?
[00:43:28] Abhay Gupta: There's a bunch of newsletters out there actually that I do subscribe to, like, you know, ad Tech Insiders is one.
Tech and learning is one. I think the A-S-U-G-S-V has like a newsletter as well that's really good to learn from. And then there's also like. At tech communities as well, like on Slack or online, where a lot of people are sharing great ideas. And so I think, I think I've learned the most really by talking to educators firsthand, like that's definitely been the most impactful for me.
So that would be my number one. But there are a lot of resources online too that I would recommend looking for and just learning about.
[00:43:57] Alex Sarlin: Yeah. Fantastic. We'll put, uh, links to any of those resources in the show notes for this episode. As always, things are moving so fast that staying in touch with other people, building community in, in the space so you can actually keep up because it is just, this whole space is moving faster than anybody can possibly, can possibly keep up with.
And as somebody who tries to keep up with it, trust me, it is so hard the Claire's A-S-U-G-S-V. AI newsletter is, is incredible resource as well. Thank you so much. This has really been really interesting. Uh, Abhay Gupta is the co-founder and CEO of Frizzle coming from the product management world at Coinbase, Tesla and Meta and Building Fr.
And he's also scaled an online e-learning platform. He is building frizzle to help teachers grade in math and beyond, in English and beyond and internationally. Thanks so much for being here with us on EdTech Insiders. Thank you so much for having me, Alex. It was a pleasure. Thanks for listening to this episode of EdTech Insiders.
If you like the podcast, remember to rate it and share it with others in the EdTech community. For those who want even more, EdTech Insider, subscribe to the free EdTech Insiders Newsletter on Substack.
Woo.
