AI in the Classroom: Practical Insights from Leading EdTech Innovators

Yay. Now, a little bit about us. AIX Education is a student-run initiative dedicated to bringing together students, educators, researchers, and founders to collectively explore the future of AI in education. We host events like this webinar and also conferences and publish a bi-weekly newsletter to share the latest discussions and research in this very exciting field.

Thanks, Sophie. In today's webinar, we'll be featuring leaders from edtech companies like Brinley and Budbox as they build AI solutions for learning. I'm excited to introduce today's panelists, starting off with Bill Salah, the CTO and CEO of Brinley. Bill has more than 20 years of experience overseeing large-scale development projects and has more than 24 years of experience in web application architecture and development. Bill founded and served as CTO of multiple internet and web development companies, leading technology projects for companies including Age of Learning, AOLL, Educational Testing Systems, Film LA, Hasbro, HBO, Highlights for Children, NBC Universal, and the US Army. Bill currently serves as the CTO and CEO of Brandy, the world's leading AI learning companion with a vision to give every student in the world access to personalized learning, no matter their background or resources.

Well, thank you so much for joining us. I'd love for you to share a little bit more about your work at Brainley and the AI platform you've been working on.

Yeah. Well, thanks for the introduction. Um, I've never actually heard that whole thing um read out like that, so I feel like I needed to tell our team to cut it down, but uh but thank you. Um yeah, so Brainley uh Brinley's an amazing um student-founded, student-led uh platform. The founders have all now grown past their their student years. Um but we come from the point of view of uh students solving problems for students. Right, so the the idea initially was um from our founder Miho; it was, "Hey, you know, I'm benefiting from study groups in my offline life, in my real-world life. Um, you know, I use these study groups to help me learn; uh my study groups greatly helps me do my homework; helps me overcome my my educational challenges." And his philosophy is like, "you know, I love these groups because nobody knows everything, but everybody knows something."

And what he found as he went into college uh in Poland uh is that these study groups weren't as prevalent as they were in high school. Like there's a really strong culture in Poland around study groups in high school, but college is a little bit different, or was at that time anyway. So he wanted to recreate that study group experience online. He created this product, and it just took off like within a year. There were millions of people using it. Um within three years there was, you know, tens of millions of people, and we've grown to hundreds of millions of people using our product now. Um and you can imagine that, you know, as this product has grown, we've evolved our our our product offering. We've evolved our point of view um on how we can answer questions, on what type of help we can offer. And we've gone from that sort of initial seed of an idea of a study group for everybody into this more full-featured educational platform.

And our mission today, uh as you mentioned, you know, we want to bring personalized learning to everybody in the world, but we don't want to replace the existing education systems. We feel that our role is to work with them and help fill the gaps. So unlike um what you might see in other places or or other edtech products that you might have come across, our goal isn't to create a secondary parallel school. Like we're not asking you to go to class and then come take a class on Brainley. Um we're building a platform that learns how you're learning in school, that learns the content that you're learning in school, and then adapts to what you need to know to be successful in the classroom. So that's our goal. There's a lot of ways that we do that. I'm sure we'll talk more about that uh later on in this in this, but I'll I'll shut up now and let you move on.

Awesome. Thank you so much for sharing, Bill. And now I'll be moving on to France Hong, co-founder and CEO of Boodleblock. From Vietnamese refugee to West Point graduate and White House lawyer, France has been pioneering the future of AI-enabled learning and collaboration in education. Under his leadership, Boodlebox has emerged as a transformative GenAI platform, serving over 10,000 faculty and students across over 650 universities. France has experience in building companies with over $600 million in combined sales, and he brings unique insights into how educational institutions can effectively harness AI to enhance learning outcomes while maintaining the essential human element in education. France, I'd love for you to share a bit more about your platform.

Great. Thanks, Lily. Um, first I want to just say, um, really appreciate being here, the chance to speak with you all, the chance to speak to all the educators and students on here. We really view what we do as a partnership. So, I'm I have an educational background. I've been a trustee of a university. I'm a distinguished visiting lecturer at West Point, so I spent plenty of time in the classroom. Um, but I'm also a technologist. Um, I've been doing, I like to joke, AI before AI was cool. And these things combine into Boodlebox, right? We we built Boodlebox because we felt there's just too much AI and AI, and we wanted to build a platform around this idea of collaborative AI, right? How can multiple people, multiple AIs work together in a responsible way?

So, um the idea behind Boodlebox is to create a collaborative AI platform for lifelong learning and work, right? and we've launched primarily focused on higher education, though we are a workforce tool and we are used by folks that are in 9 through 12 as well, but we really want to reach tomorrow's workforce today, and so that's the idea behind Boodlebox. Um, you know, there are a number of things that are being questioned about higher education now, not the least of which is what is the role, right? and what should higher education accomplish. Um, you know, I get the privilege of being able to talk to hundreds of educators and hundreds of students every month about um what's going on in higher education and AI. And I've come to the conclusion that if we want to succeed in higher education, we really need to talk about AI readiness. But what does that mean, right? Like what does it mean to be ready for an AI world? And I think it means three things. The first is you have to be a domain expert. Um, in other words, you need to be able to know enough about something in order to evaluate and explain what AI produces and whether or not it's right. I think we've all had the experience that AI is actually most valuable when we already know what we're doing, right? And so, education needs to create these domain experts.

Second, um, we need to develop the ability to recognize when and how to use AI responsibly. Um, I'm a co-founder of an AI company. I forget to use AI. Like I literally need a post-it note that says like, "Have you tried AI?" Right? And so part of education I think now is to build in um ways to teach students to recognize those opportunities and then how to do it in a responsible way. And the third thing I would say is I think in an area of AI, the things that make us uniquely human are actually more, not less important. The ability to collaborate and create and critically think and communicate. And so as we go through education, I think just merely possessing knowledge, right, um isn't as important. In fact, it could be in many cases cognitively offloaded. But what to do with that knowledge and especially how to work with other people is the important part. And so Boodlebox is the technology platform that enables all three of those things: developing domain expertise, creating this AI enablement to be able to recognize how and when to use AI responsibly, and then basically creating more excellent humans that are doing the things that humans do well, even better. And we do that by creating a platform that isn't just about teaching with AI, but teaching about AI. And I'm happy to go into more detail as we continue this conversation today.

Awesome. Thank you so much, France. I've also been a user of Boodlebox, and I can definitely see those elements coming through, like the collaborative features and how you can use different models to bring your thoughts together. So that's really cool. Um, well, thank you both for sharing, and I'm excited to start off with some questions that we prepared in advance, and during this webinar if anyone in the audience has questions they'd like to ask, feel free to put it in the chat at any time and we'll make sure to keep an eye out for those as well. So starting off, this is a question for both of you: How did you personally get into AI and education? Bill, do you want to start us off?

Sure. Um, by accident is the short answer. Um, the longer answer is, uh, just to give you a little bit of context about where I came from and how I got into technology in general. Um, I'm a person who taught myself how to write code. Um, I was not employable as an engineer for for I'd say probably more than half my career. Um, I started my first company because nobody would hire me as an engineer. Uh, if you didn't have a degree back then, you didn't know how to do the job. Um, yet here I was doing the job. Uh, writing code, uh, building websites, building applications. Um, and people were hiring me, asking me to do that for them. So, I started a company around that. And some of those clients were education companies. And I ended up working on some education initiatives for America Online. This is way back in the day. Probably a lot of you guys don't know what America Online is. It was how you connected to the internet back in the day. Um, they would send you uh physical discs in the mail that you would put in your computer that would connect you to the internet. Um, well, they evolved past then, and they were quite further evolved when I worked with them, and they had projects for education um in the classroom; like they were making this big initiative to move a whole suite of projects, products into the classroom, and some of those projects and products that I worked on for them were AI-powered. So we were we built this one product called the Writing Wizard, which launched around the same time the first Harry Potter movie launched. In fact, they co-advertised the Writing Wizard with the first Harry Potter movie, um which was cool. Um but that project was AI-powered and allowed you to write an essay and um get AI-powered feedback on how to improve your essay. There was something like 400 different things that that AI could do in terms of um helping you write a better essay. Um this is also a place where I collaborated closely with ETSs. It was the first time I worked with ETSs. Um which is a great group of people and uh and yeah, from there I kind of built a specialty in AI and education. You know, it was just completely by accident that this project landed on my lap. And you can kind of see my whole career was somewhat of a purposeful accident up to that point. The thing that shouldn't happen, but I made it happen anyway. And here we are 20-something years later um still doing it.

Awesome. Yeah, I love your journey of how like you really kind of made it happen. You started to learn how to code yourself, and here you are um now being able to create more AI platforms for students. It's really awesome to hear. Um, France, to you. How would how did you personally get into AI and education?

Yeah. So, um, as I mentioned, right, we had the idea for Boodlebox. We built it. This was this was, you know, more than a year and a half ago. Um, we wanted to build a collaborative AI platform where multiple people, multiple AIs can work together in a responsible way. We built the platform and released it, really aimed at the workforce, and we really weren't weren't sure where it was going to start and land, right? because obviously there's one of the challenges I think and strengths of AI is you can do anything with it, and for some people that empty chat window is a world of opportunity; for some people it's, "Oh my goodness, I don't know where to start." Um so we built this workforce platform, and I still remember going to West Point to teach and then talking uh to one of my colleagues, and he talked about how he was using AI in the classroom and how um he allowed on the final project students to use AI if they wanted to. They just had to disclose that they did it and how they used it. And what he found was that there were three groups of students. There were students who um didn't use AI, and their grades varied widely, right, from A's through D's. There were students who used AI and then clearly didn't do anything but take the AI output and then submitted it, and they got kind of that that solid B minus, C plus that AI gives you in that environment, right? But then there was a group that clearly used AI and then edited it and added their own expertise and human experience and judgment, and they got consistently the highest grades, which I thought was really interesting. So this is more than this is more than a year ago. And so I did a blog or about a year ago. Um, I did a blog post on LinkedIn, and it absolutely blew up, like thousands of of of of views and hundreds of comments, and a whole bunch of educators reached out, and I talked a little bit about Boodlebox, and it was like, "Oh well, this is interesting; like this really seemed to hit a nerve, and maybe there's a real urgency here to AI and education that we hadn't considered," and so we decided to go to the Teaching with AI conference in Orlando to meet more educators and more students. And there was, I think, 800 attendees from like 150 schools. And in a matter of three days, we had over a 100 demo requests from 80 schools in a matter of like, you know, 3 days of being on site. And so that's when we realized that there was a really urgent need here. Um, it it met what our our solution met the kind of problems that that education was facing. It was something we were already passionate about. I spent lots of time in the classroom, um and so we decided to focus on where tomorrow's workers are today, like I said, and focus on the classroom, right? and and um working with professors and with students, and very shortly thereafter started a collaborative AI leadership council consisting of professors and academics and administrators as well as a campus student AI leadership council to get the view of students, and that was um it's only been actually seven months since then, and so we've been off to the races ever since.

Awesome. I think especially with startups, like you don't know where you're going to end up, but it's so cool that you're able to narrow down on that niche towards AI and education. So, I'd love to learn more about how does AI power your specific platform? What do you think makes your approach unique? Starting with Bill.

Um, a little bit of a story again. Sorry, I'm a storyteller. Um, okay. So the way the AI was at Brainley before I was at Brainley, and what's interesting is when I joined the company um I was coming out of an environment where we were very heavily invested and excited about AI, another education company. I came into Brainley, and we were we were not excited about AI uh as a company, and one of the things that I was tasked with was to sort this out, like figure out what we're doing here. What is our what does this investment mean to us? and like if we can't make this make sense then kill this investment because this isn't really going anywhere. It's been a couple of years. Um so I was super bullish on AI coming in, and I was like, "Hey, wait a minute, this is probably one of the most important areas that we can invest in. Why isn't this working?" Um and we really had to go back to first principles to make AI work for us. We had to understand like what is our market fit? What type what problem are we solving? um how can we solve it better? Just to get to the point where we were like, "Okay, now that we know that about ourselves and what we want to do and who we want to be, how can AI accelerate our outcomes?" Um and so we started this at the time which uh started this five-year investment at that time, right? We were like, "Okay, five years of work here; it's an AI-based road map, but not because AI is cool, but because AI will actually help us achieve what we want to achieve." This is pre-ChatGPT days by the way, so everybody has context. Like AI and education has been going through these cycles of like cool then hot, cool then hot um off and on, and like everybody would throw labels on their product that their product was, you know, adaptive learning and personalized learning, and like it really didn't mean much in in edtech. Um it was kind of like one of those superficial labels. Um and everybody had a similar approach. So like we came up with a really novel approach, and that's the answer to your question. How do we use it? We use AI because we have hundreds of millions of people using our product. We use AI to sort through that data and look for patterns. So what we're trying to do is understand like what are you learning? And if we look at our corpus of data, our entire knowledge base, we can actually recreate curriculums by watching people traverse our data sets. Right? So if we see you accessing content in a certain pattern and certain pieces of content, we can actually build cohorts who are doing similar things, cohorts of people acting similarly. And we can then start to create predictive um interventions and also reactive interventions. And this is a way to say like we can look back and understand why you're struggling with something today based on your history of usage of Brainley. And we can also predict what you're going to struggle with in two weeks, two months, etc. based on what other people in your cohort who've gone through the same curriculum a year before you struggled with. Like people who act like you tend to struggle like in these ways in the future. It's kind of like the, you know, the same ideas behind how Amazon sells you more stuff or Netflix gets you to watch more videos or YouTube um gets you to watch more content, right? So those types of of known problems and known solutions, we're able to apply to our data by taking an approach of um learning the curriculum dynamically instead of pre-creating the curriculum and instrumenting it and forcing people into these these predefined instrumented paths. So AI does all of that for us. There's a lot of of technology involved in making this work. Um, and then we most recently are using all of these systems to go a step further. So we have on Brainley, this is this is an experience that not everybody will get today because it's currently only randomly available to 20% of the people who go to brainley.com today on the web. Not available on our mobile app yet and not available in any other um global market. But 20% of the people in the US today might encounter an experience where they're actually interacting with this thing we call the learning companion. And the learning companion is a really advanced tutor that tries to learn about you. It tries to learn about your struggles in the subject that you're working on. And then it tries to learn about your homework needs. And then down at the lowest level tries to learn about why are you asking me this question right now? And it takes those four levels of context and that larger system that I described. And it creates a very personalized response to try to unblock you. So this is where I go back to saying like we're not trying to teach it differently than your teacher or give you a parallel school system. What we're trying to do is understand like why are you asking us this question? What are you struggling with? And how can we best help you get unblocked so you can move forward and have better outcomes in the classroom. Lots and lots of technology involved here, lots of systems that I'm kind of glossing over. Um, but that's the role of AI for us today.

Definitely. I feel like in the past when we had like people like human tutors, sometimes it would take a long time for them to best understand their students, understand where they're struggling. But with AI, it can do it really fast, and you can really understand those patterns. I love how the AI tool really goes down to the root cause, like what are you really struggling with, and that can really open doors to so many more opportunities. Thank you so much. Um, and France, can you share a bit more about how AI powers Boodlebox as well?

Yeah. Let me start by just talking about education, right? Like so for hundreds of years we've probably educated in a very similar way, right? uh you go to class, you're given some materials, and then you you study that uh and then you have to demonstrate your comprehension and critical thinking by producing some of something called an artifact of learning, right? And that artifact would typically be take a quiz or produce a project or write an essay, and that's and then you get feedback on that, right? The professor assesses and gives you feedback, and that's the traditional path of learning. Um, AI has completely disrupted that because what used to be the artifact of learning that would demonstrate critical thinking, taking the quiz or writing the essay or producing the project, can now be done by AI. And so there's been an outcry among many educators, right, that AI has disrupted the educational process. That AI is now cheating because students don't have to understand anything or critically think. Um because of that, right, there's been a lot of educators who who have been, you know, advocating for rethinking education, and frankly, this predates even AI, that AI should that education should never have actually been about just kind of consuming content and then producing an artifact of learning, that AI should, you know, that in an era of AI we shouldn't focus on the product; we should focus on the process, and that the process matters; the product isn't your score on the quiz; it's not the essay; it's not the project; it's how you get there. Because in today's day and age, I joke, but it's not a joke. We are now all a B minus at everything. Like I could I I've asked you, you Lily or you Sophie or you Bill to produce a podcast script in Swahili about refrigerator repair, and you could do it despite the fact that you're not a podcast host, maybe you two are now officially, because of this, normally or know anything about refrigerator repair or even speak Swahili. And so clearly it's not the product that matters; it's the process. And so um why am I bringing this up? Because in an era of AI where process matters more than product, teaching should be about process and about learning how to use the tools responsibly. But in order to do that, then educators need to be able to see the process by which students engage with AI. And that's the difference between teaching with AI where that process is all hidden in the background and teaching about AI where it's transparent in the open. And so in Boodlebox, we're a platform for for learning and teaching and working with AI across a lifetime. And we make this transparent because of our collaborative nature. So in Boodlebox, we combine all the top AIs in one place. Whether you're a fan of ChatGPT or Claude or Gemini, they're all here. What's interesting is you can use multiple AIs in the same chat, which is really good for teaching, right? You can ask ChatGPT a question, then you can ask Sonnet how to respond. So this is teaching students the difference between AIs, and then you can share that with a professor. So a student could do work with AI and then share a link to their professor, and when the professor clicks on that link, they can see all the students' work. They can see what AI they chose, what prompts they used, what responses, what was their train of thought, and that doesn't apply just to an individual but also to groups. So in Boodlebox, you could have multiple students talking to each other and talking to multiple AIs and doing work. I would argue this is the future of work and

We enable that in a way that is transparent to the professor. And then last, and then, and I'll shut up myself.

You can build your own custom AIs. This is really interesting for building things like personalized tutors, but also simulations, but also you could build bots that then go into these group collaborations. So, um, there's a there's a number of professors who building the bot is now the class. Like, instead of taking a marketing class, you take a marketing class, you build a marketing bot, and then you solve a real-world problem with that marketing bot in a group. So you're learning marketing, you're learning bot building, and you're learning how to collaborate as part of a human-AI team, which I think is a fantastic evolution of education in this era of AI. Wow. I really like the whole idea of building the bot because you really had to think about like what should we teach the bot? How do we make sure it's the most effective?

Um, when I worked on a video editing project with a group of people, we actually used Google Fox. Um, we brought in a video editing bot to help us with the process. And what's great about the platform is that you can also you guys can work on different models together and see the outputs together, and you can also communicate, like message on the same um chat. So it really fostered that collaboration that you're mentioning. Awesome.

And as you guys have been building these AI tools, do you have any surprising insights or challenges that you faced in building these AI solutions? We can start with Phil.

Yeah, sure. So I mean I I mentioned this a little bit already. We had to go back to first principles and really understand our business to be able to make AI work for us. Um, but then when we're as we're building solutions using AI, there's a number of interesting findings that we have and and it it keeps coming back to these first principles like are we trying to recreate an offline process online and is it better than what you can do offline? Like I'll give you an example just literally just happened in the last week or two. um, part of this platform that we're building to help students do homework uh with this learning companion as a as a you know uh an always available tutor. This learning companion is able to see the work that they're doing on our platform and and give assistance, right? And when it comes to doing math homework, um we want to be able to let the the learning companion see the step-by-step solving process. Like teachers want students to be able to go through the process to get the right answer. They don't just want to see the right answer. Um, and the learning companion needs to see the student go through the process so that it can coach them, it can assist them. Well, it's not a very straightforward process to just take what you do on paper and put it in a digital platform. Like, if we if we were to just recreate what you can do with a with paper and pencil, we wouldn't be making this easier for the student. And this is something we constantly have to remind ourselves like we have two bottom lines here in what we're trying to do. We have make it better for the student meaning it should be easier, more efficient, more effective to do your homework on our platform, but also for the teacher who's going to be grading that homework and reviewing and auditing the tutoring session. We want that teacher to trust our platform and see that the student got a great educational experience. So, like, how do you hit both those bottom lines in a product when it's math? And frankly, the best ma best math solutions that you'll see today are these really cryptic keyboards for like entering special characters and you're hunting and pecking with your mouse. Like, that's just not easier than paper. And we could have sort of cheated and said, well, what if you just do it on paper and then you hold it up to the camera? Well, again, that's not easier than doing it on paper. We're just getting in the way. So like we're constantly reminding ourselves that this double bottom line exists and to use our platform for students to really love it. We have to make it easier for them to use it than to do it in a traditional way. And then we can layer on top all of the AI magic. So the AI magic there is it can't be the only value proposition for our students, right? Otherwise they're just not going to use us.

Definitely. I think as a student myself, it's like sometimes like I people it's really easy to just put like short prompts into chut because it's a lot of work trying to write everything or even like doing math on chbt trying to get the right characters is really difficult. So making that seamless and ease is going to be very important for sure. Um, and next France.

Yeah. So um lots of surprises. Um, you're always learning. Um, like I said, we view what we do as a partnership um with students and faculty administrators. So we get lots and lots of feedback. And one of the things we realize is we're actually not solving problems for one group, it's multiple groups. So for the students, one of the core problems that students have is this question of equitable access. So if you use better AI, you get better outcomes, right? And this is a problem, right? Like we an outcome in a classroom should not depend on whether you're using a free tool or whether you have the money to pay for a tool. So there's a number of institutions and professors who are very concerned about this and are looking for a solution to provide equitable access to AI across their entire institution. And we're in talks with dozens of institutions now who are interested in come the fall when a student shows up to class they all have access to the same suite of AI tools of premium AI tools. So there's so everyone can start from a level ground. Um, on the professor side um I think professors are largely through their five stages of grief with regards to AI in the classroom and that if not willing to embrace it at least accept that it's here but they want to know where to get started and so um in budbox we have provided a lot of resources that are geared towards educators. So if you go into Boodlebox um depending on when you sign up you're actually asked are you a student are you administrator are you an educator and you have a different interface in different resources so if you're an educator you get this interface you have a coach here that will help you use AI um more efficiently give you ideas if you've never used AI before and you have this wonderful resource page that has dedicated resources aimed at educators for building assignments ments, preparing for class, teaching and learning, assessing work, as well as links to various bots and tools that are geared towards higher education. And then finally, um you know, we always have to remember that there's also the institution and administrators and they have to deal with things like is the system secure, is it private, is it accessible? And because we focus on higher education, um we've checked all those blocks, right? we have hecvats and vats and we don't train on models and you know we have a sock 2 compliance and we have this really affordable pricing model um because one of the um innovations we came up with was we figured out how to run um large language models such that we use up to 80% less tokens per conversation. So it's 80% less energy usage that's 80% less impact on the environment. It's also 80% less costs. So we're able to provide unlimited access to all these AI models at a price point that's actually lower than a single subscription any one model. Um which solves again that problem of affordability for the institution but also equitable access to the students. So we're constantly finding these new friction points for the students for the professors and for the institution and trying to solve those.

Definitely. I think equitable access is so important and I really like how Box also focuses on the environmental aspect as well. I feel like that often gets unnoticed even though it does have a significant impact on the environment.

Um, and for your guys's solutions, how do you measure the success or success in learning for students? Starting with

Yeah, that's a I mean there's there's an easy answer to that, but I don't think it's the best answer. Um, especially because we're a student-founded company and we really take the perspective, like I mentioned in my last answer, that it's got to work for the student. they have to love us. So, I mean, the the sort of de facto go-to that I've had throughout my career working with edtech and and especially AI and edtech is you do independent third-party reviews of your product. So, you you put it in classrooms um you measure the impact and there's tons and tons of of um examples of this in my career and it works. like it's an awesome way to figure out like are we getting the educational impacts that we expect out of this tool, but I think that's a really small part of the problem in terms of the overall, you know, thing that you need to to solve for. Um, if the teachers don't love it, if the students don't love it, the impacts won't be made. You know, the fact that the tool works is kind of uh table stakes. Just building an edtech product and it not working like you don't have a company, right? So you have to build something that works, but for it to be successful, you really need people to love your product and you have to be solving problems for them. It has to be easier to do your homework on Brinley than to do your homework without Brinley. It has to be a better experience for the teacher to be able to um give homework on Brainley, to grade homework on Brainley, to be able to audit those tutoring sessions, to get real-time feedback about student performance, right? um to be able to see like, oh, I I I gave this um you know, I gave this lesson yesterday, gave the homework assignment, and then that night if they wanted to, they could log in and get a report from Brandley that says, "Here's the topics that are trending with the learning companion tonight in your classroom." Like these things are super super important for people to love your products. Um so I think that's like the bigger answer. Uh how do we measure that? Um it's sort of traditional product metrics. It's about engagement. We do user surveys, we do focus groups, we do a bunch of lab work where we do innovative new things and we release it to smaller groups and we try to get really close working relationships with them to understand if we've solved the problem better uh where we could improve. And then like I said, you know, the table stakes is hiring those third-party independent companies to come in and do those studies as to whether or not your product actually improves learning outcomes. And you've got to nail that.

Definitely. I remember um back in middle school, high school, and even college when you get homework, it's kind of frustrating when you get to a point you're like, I don't know how to answer this. Like, I wish I had someone to help me like a learning companion. So, I could see Brinley being very effective in being that um tutor alongside these students when they're doing homework where, you know, they also give you independence to really figure it out and struggle through on your own, but have give you that assistance when you need it.

For sure. Um, in France, what about you? How do you measure success for Brut?

Yeah, there's a lot of uh quantifiable measures, right? Like what percentage of people sign on when they're invited? It's above 70% in the first three months, right? What percentage of people are active users? It's about the same amount. And it actually gets stickier over time, right? Um, what percentage of students are um actively using it for work? What percentage, you know, how many bots are being created by professors? and all those measures, you know, we pay lots of close attention to those and we provide reports to the institutions that um purchase Google Box and and show those measures of activity, but ultimately that's just activity, right? That's not what we're looking for, right? The true productivity. And so we have a lot of conversations with with users, students, and professors and get a lot of subjective, right, qualitative data about uh budbox and and its effectiveness. But the ultimate measure, right, is one I can't measure yet. And that's because it hasn't we haven't had the chance to see it. So our ultimate vision for Boodlebox is this again this lifelong AI tool. What do I mean by that? It means that you know Lily or Sophie, imagine if you showed up in your first day of class and you got access to Boodlebox and for the next four years, every chat, every bot, every lesson, every group project goes in there and instead of and when you graduate, instead of having a pile of books in the corner and a portfolio that is static, you have this AI toolbox that you spent four years building and that you can show an employer and say, "Look, this is what I can do with AI." But it's not just this dead toolbox, right? That that end of your educational journey becomes the start of your professional journey. And it's a toolbox you can use for life. You know, every skill, every bot you built, every conversation you had, you can pull it out for the rest of your professional career. And we think that's an amazing gift that an institution can give a graduate. I think it's an amazing start to a professional journey. Um, and I I think it is something that um will increase employment, right? um ultimately, but I also think it's a way where as we talk about AI, right, what we're often talking about is offloading the things that are drudgery so we can spend more time in the things that are human, right? And I've man, if I had that tool now, right, if I had spent years loading up blue box that way, I could spend more time in the things that are really important. And that ultimately is going to be the measure of our effectiveness.

Definitely. I love your vision, France, about it being like lifelong. It doesn't just stop in the classroom, you can use it beyond in your internships, in your jobs. Having a portfolio to share that keeps growing with you, I think is really effective.

Um, and what strategies do you guys have for effectively integrating those platforms whether it's Brainley or Brutal Box into the classroom or for students um in order to enhance teaching and learning?

Yeah. So, um, we have a a sort of multi-part strategy. So we are direct to student that's been our traditional distribution model. Students we have positioned ourselves as sort of this rebellious product that lives outside of the classroom for our entire existence. Um, only now are we even thinking about how to build stuff that works on both sides, the student side and the teacher side. And this was an evolution for us as a product and as a company uh because we're we're thinking about this problem more holistically. you know before um we started this big AI investment which is five years ago now that we started it um we were thinking about how to get the best explanations on our platform so that students could find information. It was really like a poll model like students come and they they look for information that would help them get unblocked. we over the time evolved into this let's understand why you're asking what you're asking and why you're asking and instead of you looking for it we'll look for it and we'll get you the best information. Um and then that evolved into a hey information is just part of it. like we've built this really robust and and um sophisticated system to understand what you're struggling with like this is essentially a tutor and so if we are going to build a tutor who is the customer for that and it I think it would have been easy for us and maybe a little bit lazy if we said well it's just for students but we looked at it more holistically and we said well you know for it to work well for students it has to work well for teachers also it can't be something that we put out there and we think about um this tutor not being aligned with how the teacher's teaching. So the first version of our tutor that we did in a lab and that we shared with teachers and with students, it actually was configurable completely by the teacher. So the teacher chose the pedagogical approach that the tutor would have. Did it want it to make it more didactic? Did you want to make it more socratic? Like how would it respond to certain questions the student was answering? And the feedback that we got was amazing. like teachers were like, "Hey, you you know, for this particular student, I would want it to work this way, and for this other student, I would want it to work a different way." And we had, we hadn't even thought about that level of customization. Um, which was a super interesting find for us. And that really opened the door to be thinking about, all right, maybe this should be something we put in the classroom. Maybe this is something that we should turn into a larger product offering. So, our approach now to to working with educators is still very much like we're not trying to sell to them. We're we're currently not offering a product that you can buy and put in your classroom. What we're trying to do is build a product that students and teachers love. And when we hit that goal, when we see that organic sort of grassroots swell of support for this thing that we've built, then we'll be able to think about, okay, you know, x percentage of teachers in this district are using it. X percentage of students in this district are using it. Here's our, you know, independent research study saying that it improves educational outcomes. Now, maybe we approach that school district and we say, "This is already happening. you know, here's some more things that we can unlock if we wanted to to open this up to you as a a tool that your district uses. Um, so that's far off for us. That's in our future. Right now, we're just all about providing everything we can for free and building that organic ground swell of like a product that people love to use. selfishly and and we're not doing this completely um completely philanthropically like selfishly we want people to use it and tell us how to make it better right like that collaboration has to happen before we can even think about trying to monetize it into the school systems when we get there um I would expect that there's a lot of hurdles um France I'm sure you you you've had tons of these challenges we know what the sales cycle looks like into schools it's it's a very difficult thing Um, and frankly like we're just not ready to even approach that. We're still on the just tell us how to make it better mode of Brley.

Yeah, definitely. I think getting that user feedback is really helpful to improving the product and I can see it like later in the classroom at least in the college setting. um using AI like the learning companion as somewhat like a TA would be really helpful within the college like um the TA is only available maybe like once a week for one hour and if you can't make it then well good luck you know but I think having that learning companion available in the classroom as kind of a support for teachers can also be pretty helpful.

Thanks for sharing Bel.

Yeah. So, um, AI is really interesting, right? Because here's here's kind of the dirty secret AI. It's the students that drove this change, right? It's not like universities woke up one day and says, "You know what? We should get we should incorporate AI, right? Or it's like a bunch of prof there was a few professors were excited about it." But the students started adopting this in droves. I mean, the last latest stats were like 80 90% of students using AI. And so the students forced this change and now the professors some of them were excited about this because they were excited about change anyways right other professors like I said had to co overcome their five stages of grief and they're at a place where they want to engage with AI because they see the pedagogical value of it and then institutions I think have a need to both meet this requirement for equitable access but they also want to be relevant right and are concerned about producing graduates who are going going to be employable and prepared for an AI-enabled world. Um, so we're fortunate in each of those places we can interact. So we have the ability for individual students or professors to buy individual licenses and we have plenty of those. Um, thousands of those. Um, we have teams so you can purchase it as a team. A professor can buy it for their classroom. Um, we've had groups of students buy it for group projects. Um, we've had schools use it. And we have this pricing model which um is based on activity in that case. Now what does that mean? It means you can give access to everybody and then every month on a month-to-month basis you just pay for the number of users who are active on the platform because as it turns out one of the challenges of AI is you can't adopt what you don't have access to but you don't know how many people are actually going to use it which is a real catch-22 to providing edtech technology right and then at the institution level we do offer institutional site licenses. We're in discussions with 30 institutions right now about that. um to provide complete equal access to to everybody. Um we are seeing um some really really novel use cases um by institutions that are kind of our early adopter. So I'll I'll put one out there. Um tutoring, right? It is um one-on-one human tutoring is a very time-intensive activity. So there's a professor at West Point who now requires he calls it AI for AI. before you go for additional instruction AI with him, you have to use his artificial intelligence bot and engage in a conversation and try to help yourself with the problem. And then if you want AI in person, you actually send him a link to that conversation you had with the AI bot. Um, and when you show up, then a couple things have happened. One is the easy questions have been answered by the AI bot. you've done some work on your own and the professor can see all your work so that they can spend their time on probably that 20% of one-on-one tutoring which is actually really critical right helping a student overcome a mental block encouraging them you know the real teaching whereas the 80% of work that could be done by AI was done and I think that's a wonderful example of this kind of human-AI collaboration right there's teaching with AI there's teaching about AI there's teaching by AI right and I think you know uh and it sounds like you Bill doesn't believe in this either, right? Like you're not trying to replace the this teacher, right? We want to augment, right? And we want to enhance. And I think that's a great example of that.

Awesome. Thank you so much. I really enjoy having this conversation just getting to learn more about the leaders behind these ed tech companies because like, you know, I use it, I interact with them, but never really thought about things like the pricing model and also the challenges that you guys face. So, this is really interesting. Now I'll be tossing over to Sophie to ask any additional questions and we're happy to take any questions from the chat in the audience as well.

Thank you Lily and thank you Bill and friends for sharing all these amazing insights from all the hard work you've put in.

Um, I have I want to start with one follow-up question um for Bill. I love this vision that you painted for how you can we can leverage AI for personal learning to fill in the gap of um traditional classroom learning. And this is a vision that a lot of um act companies, a lot of educators and innovators have like brought up over and over again. Um, it goes back to learning analytics from like decades ago. And now we have a lot of different learning companion products that also share the same vision because this is such a great role to try to pursue. I don't know if you've um there's like always one dilemma. We know that productive learning is hard and making a tool for students to use that actually like intrigue them enough to want to pick it up and use it. That's implying that you need to facilitate productive learning. Um that is more interesting than student just ask

Say “JP” for an answer. How are you currently facilitating this, like, design or learner experience to make sure that studying with Brainly is more appealing than just getting the answer from other AI tools?

Yeah, it's a great question. So, first of all, I'm going to say there's a lot of room for improvement here. Um, I don't think we've nailed it, and, and here again, most of the answers I've given today are specific about the learning companion, which is this whole set of homework solutions that we're rolling out that, um, that, as I mentioned, are only available to 20% of the people. It will be more—it's a slow roll-out because there's a lot of newness in what we're building, and we don't, we don't know what's going to break. Um, so we're kind of, you know, testing that.

Um, but back to your question. So, there's a couple ways that we're trying to solve this problem. We are trying to connect everything you're doing on Brainly to everything else you're going to do on Brainly. And this goes back to like what I talked about earlier. Just by looking at the data that you're accessing and then building cohorts of other users who are accessing data, we're learning how to help you. So, we have these built-in mechanisms just by the scale of our data and by our use case.

Um, one of the things that I had to remind our team is that look, every page or every question that somebody looks at on Brainly is an indicator that they needed help with that topic, because if they knew the answer, it wouldn't be easier to come to Brainly than to just put the answer down without searching Brainly. So, people are coming to Brainly because they don't have the answer already. So our job here is to give them really high-quality explanations.

As we moved into these homework products, we started thinking about, all right, we've seen your, your, your content visits on Brainly, we've seen the data that you've accessed. How can we leverage this in these homework products? So like an example of how we make this easier is that we have a test prep product that we can actually build questions to help you practice for your upcoming test based on what we know you visited on Brainly. If you want to upload a document, great. If you've done your—to that we can parse and pull, you know, build questions from, great, we can do that as well. If you've done your homework on Brainly using our, our homework worksheet product or feature, we can actually let you link your homework to your test prep creation. So now we're looking at the homework that you've done, the conversations you've had with the learning companion while you're doing your homework, and we're building practice questions off that. So you can see just in the design of the feature, we're building a practice test around what you need to know, not around some generic concept of what you're being tested on. And then when you take the first steps, we adapt the second test to your performance on the first test and so on and so forth. We actually build out a, a study plan for you where you take an, a series of practice tests and each one adapting to what you did the last time and, and focusing on the areas that you need more practice in. Um, things like this that make learning more efficient for you. It's about personalization, not necessarily just that challenge and response.

Now, challenge and response is important. And I'm going to, I'm going to end it by, by sort of giving this, this analogy that I've given or this explanation that was given to me by, um, somebody I highly respect in the education world. This is a consultant that we work with. He's awesome. Um, he's taught me a lot. But he, you know, we were, we were talking about learning and I asked him the question like, what is learning and how do we know when it's happening? This was actually a research project that we ran at Brainly. Just kind of like a weird question like, what is it and how do you know it's happening? What we settled on, and this is from him, is that imagine learning is the process of doing a puzzle. Everything you know are the pieces that you've connected together. You're starting to form a picture. Everything you don't know are the pieces off to the side. Learning happens when you take a piece out of the pile of things you don't know and you connect it to the picture of things you do know. So the picture gets clearer and you're connecting one concept that you don't know to a bunch of concepts that you do know to create that clearer picture. So you can see here like in no part of this analogy did I say that we're going to hide the pieces and you've got to go hunt and peck and find them, right? We're not, we're not hiding them from you. Access to high-quality information is part of the learning process. So, the tutor explaining things to you, um, and helping you connect that information you don't know to what you do know is crucial. Some people may say, well, the easy, the answers are coming too easy. But that's what our, our, our investment in our technology is all about, like getting you the information you need to know more quickly and helping you connect it to what you already know so we can help you learn. But we're not trying to hide the pieces from you. We're not trying to, to, to say, well, why don't you go open a book and, you know, search through it and like try to find that information for yourself, right? Like we want to get it to you as quickly as possible and then answer the question as to how can we help you connect this to what you know.

I love that. I have two really good takeaways from what you just said, wise words. Um, the first is we are designing at tech tools based on kind of the social context students are in. What's their job to be done and what are their goals and try to kind of design something that fits into their current ecosystem. And I also love, um, the puzzle quote, how we, how a lot of personalized learning, like a lot of personalization kind of goes back to how do you leverage what the assets students bring in themselves and how we leverage that for more productive learning. Thank you so much. I, I love the answer.

And my second follow-up question along the line of kind of designing for the ecosystem that our learners already in for friends is you mentioned we want to leverage AI for kind of future of work as well. Um, we want to make sure that the skill sets students are building, um, in terms of their AI readiness can be brought into their future workplace. So what do you think? What about AI? Um, are you trying to facilitate this learning in to make sure that they're prepared for this future of work, which we are not really sure what's going to look like? We all of us kind of have our own theory.

Um, let me try to like condense the question a little bit. So what do you think as you design a product are things that students need to learn to be prepared for their future workforce?

Yeah. So just some context, Microsoft LinkedIn did a study, um, they talked to 30,000 global employers. 67% employers said that in the future they're not going to hire an employee that does not have AI skills. And 71% of employers said they were, they would hire somebody who's less experienced with AI skills over somebody who was more experienced and did not have AI skills. So this is absolutely critical for the 18 million students out there who are going to be entering the workforce in the next four years. Right? So I, you know, I touched upon this earlier at the beginning, right? I think the three things we've got to develop, right, in as students and for students and with students, right? The first is this domain expertise. AI is best when you already know what you're doing. So you have to be better than the AI. So whether you're going to be an accounting major or a finance major or an English major, you have to be better than the AI. Otherwise, why are you there, right? So you got to become a domain expert. Second, you've got to find those opportunities to use AI to be faster, better, more efficient. There's that BCG study. People use AI, you know, on, on average are like 25% more productive and 40% faster, something like that. And so we don't have that muscle memory now, right? Like, like I said, I'm an AI founder. I forget to use AI. So just learning that to recognize and spot those opportunities on when and how to use AI responsibly. And I think as a, as a recent graduate, if you show up at work and you're like, we can use AI here and we can use it here and let me show you how, you are instantly value-add. And then third, right, just continuing to develop those most human of skills, right? The communication and the collaboration and the critical thinking that AI can't do. If you graduate with those three things, domain expertise, being AI enabled to recognize when and how to use AI, and those critically human skills, I think you are well prepared, right, for a world of AI.

Thank you. I love how you painted a very clear pathway for what students can do to be AI ready and getting into data is definitely really helpful for us to sense this sense of urgency. Thank you so much. And I will guess.

Thank you so much for both of your guys' response today. Um, we really appreciate all of your insights and your time for joining us. Um, thank you to all of our audience members for joining as well. We hope you also gathered some helpful insights through this conversation. So, today it wraps up our, the end of our webinar and we also will make sure to send a recording of this webinar in our newsletter as well so that you can follow up and review the webinar. Thank you all for joining and we hope you have a great rest of your week.

Thank you. Thank you.