After returning from a family trip to Legoland earlier this month and having some eye-opening interviews with edtech founders, I’ve been thinking a lot about an aspect of AI education tooling that I think has gotten far too little attention: characters.
Let me start with a hypothesis: the days in which the primary interface for AI is an LLM “chatbot” — a neutral, faceless, text-centered experience two steps away from a conversation with an ATM — are nearly history. Instead, with the rapid advancements in video, audio, animation, and AI memory, we are soon going to start seeing AI characters designed with education in mind, which already exist outside education in companionship apps, social media, and more.
Depending on your point of view, this may or may not be a dystopian vision. But I’m increasingly convinced it’s coming. The way it will come? Someone will have the confidence to be a first mover (Teach Me Elmo?) and they’ll catch all the flack and thinkpieces. But, it will work… culturally, financially, and hopefully educationally. Once this occurs, the floodgates are open. So let’s get ahead of it before we’re all surprised!
Learning from Lego
Even though Lego may be best known for construction kits, the company has a particularly brilliant, three-part philosophy on characters that makes an instructive metaphor as we think about AI-enabled characters.
Approach 1: Customizable Characters
Lego’s first stop in character development is their wildly popular Lego Minifigures: this product uses Lego’s classic approach in providing huge numbers of interchangeable, modular parts that allow kids to use their agency to create and adapt characters with millions of possibilities. Minifigures have become a subculture for kids and tweens. They provide Lego fans with autonomy and trigger their imagination. At Legoland, you can even walk up to any Legoland employee and trade a Minifigure in your hand with one on their badge.
Full Customization for AI Companions
Like Lego’s Minifigures, AI companion apps like Replika, Anima, Kindroid or Nomi tend to offer fully customizable AI looks and personalities that engage users’ imagination. You can tweak your AI companion’s hair color, personality traits, and a million other factors.
Likewise, some of the first attempts to inject character into AI usage for education also tended towards a modular, customizable approach. In Tolan: Alien Best Friend, users can choose the gender, clothes, and hairstyle for their adorable aliens and then chat with them about any topic under the sun:
But with AI, for the first time, even photorealistic characters are customizable: you can change the accent, language, hair color, clothing, gender, race or height of a realistic person, alongside their personality and backstory. Here’s a video of my HeyGen Avatar, tuned to be more enthusiastic than usual:
For a deeper glimpse into realistic approaches to character-driven companions and teacher characters, check out the competitive advancements in avatar technology from Heygen, Synthesia (which just launched hand gestures, and is about to drop a big 3.0 upgrade on October 1st), Google Veo, Collosyan, and Dreamina.
Approach 2: Licensing Popular Characters
Over the last decade, The Lego Group has become a world-class expert in intellectual property licensing. Walk into a Lego store (or watch a Lego film) and you’ll see Lego versions of many of the most popular and valuable characters in the world. They have Marvel characters like Spider-Man and The X-Men, and DC characters like Superman and Batman. They have Star Wars and Transformers, Minecraft and Fortnite, Super Mario and Sonic The Hedgehog, along with Harry Potter, Bluey, Peppa Pig, OnePiece, Jurassic World, The Simpsons… the list goes on and on. Like Netflix or Xbox, Lego has successfully made itself into a platform on which the largest media companies in the world, normally competitors, want to play alongside one another.
AI Licensing
There’s something undeniably exciting about repurposing characters that we have learned to love and recruiting them to support educational endeavors. Want to learn physics from Spider-Man? Meditation from Yoda? Astronomy from Buzz Lightyear? Forensics from Sherlock Holmes? On the realistic side, who wouldn’t want to try, say, Professor McGonagall from Hogwarts as a history teacher, or Mr. Miyagi as a gym coach?
There’s no question that every major media company has been batting around the idea of interactive AI versions of their major characters, but so far, none have pulled the trigger publicly. Why? Well, it should be noted that there’s a pretty major mismatch between established intellectual property and the unpredictable nature of AI companions. When I asked Brainpop’s then-CEO, Scott Kirkpatrick, about whether we would see an “Ask Tim and Moby” feature:
“Can students talk with Tim and Moby? That is clearly a potential opportunity for us that we are also extremely cautious about… as much as we believe in agency, we also have to make sure that if Tim and Moby are speaking with kids, we know exactly what they're saying, and teachers know exactly what they're saying. So there's a lot of work that we're doing to test that. Before we ever put that into our product, it will be significantly tested with significant numbers of teachers to make sure (that there are no) hallucinations, because the stakes for us are so high. In this world of AI, brands matter. And we take that very seriously.” - Scott Kirkpatrick
He’s right. The first time someone releases a “Talk to Moby” feature (let alone a “Talk to Spidey” or “Talk to Mario”* app), the countdown begins before a kid (or, more realistically, activist parent or journalist) tries to make it say something incorrect or, worse, inappropriate.
But it’s not impossible, not by a long shot. The entrepreneurial team behind the life skills app Nurture have been signing deals with established intellectual properties for kids and finding ways to incorporate established characters into AI properties in ways that don’t risk the brands. Others will, too.
*It should also be noted here that some famous characters are actually more icons than true characters. Personally, I wouldn’t want to have a beer, or even a long chat, with Mario, Mickey Mouse, or even Duo the Owl.
Approach 3: Creating AI-Native Characters
Finally, the Lego Group has made some headway over the last ten years in developing its own proprietary characters and intellectual property. No interchangeable parts here, these are characters with actual names, backstories, personalities and (celebrity) voices. They’re then featured in multi-channel media properties: movies, games, television shows, comics, and more. For examples, see: The Lego Movie ($470.8 million worldwide gross), Lego Ninjago Movie, Lego Friends, Lego Dreamzzz and more.
AI-Native Characters
In many ways, this is the most interesting approach to character development of all: the development of characters that are made in and for the Age of AI. These are characters that would be designed from the outset for interactivity, designed to balance customization and consistency, designed to be fun and flexible but also safe and appropriate no matter what the user asks of them. AI Native characters could be designed to mirror the moods of their user, or to have off days, to laugh off insults or take them personally. It’s a type of character design we’re simply not used to, a character that has to be ready for anything out in the real world, not just in the universe of their origin.
This may not be as tall an order as it sounds. Large language models are designed to follow system prompts and learn from datasets, giving a model very clear motivations and restrictions and a character bible to follow. This, along with a (long) list of topics to avoid and approach, could go a long way in making a character come alive… someday soon.
For now, we’re still in the early innings. Here’s Miko 3, an AI-Native robot. You’ll notice that while Miko is technically a character, it is a character with very little personality of its own. It tells jokes and can dance, but mostly acts a whole lot like past electronic toys like Teddy Ruxpin or Furby. Even if these early experiments were/are off-putting, character-based AI seems inevitable.
According to Ivan Crewkov, the CEO and founder of Buddy.ai, the popular learning companion-based language learning app on the Edtech Insiders podcast:
“Buddy.. is a AI native character. We're building IP specifically for this GenAI World. We should be excited about the new generation of IP born in this wonderful new Gen AI universe. It’s an opportunity for storytellers, like every new technological shift. Remember all of those great new games created for smartphones?
How do we build guardrails? Our approach is not to try build guardrails, but to… build a less sophisticated and versatile model, but a model that is hundred percent focused on the curriculum, completely free of dangerous material. And same with our visuals today. Buddy’s AI controls which asset to use.. but we still have a pre-generated library of assets so a child (can)not ask Buddy to create something weird, scary, or inappropriate.
So we believe that AI companions for children should be strictly educational. They should be built around curriculum and should be stripped of everything else. We don't want Buddy to talk about politics, right? We also don't expect Buddy to write an essay for you or help you with homework. Our approach to building AI assistance for children is to make it strictly educational and allow the building of a social relationship only as a part of the curriculum, only to make learning more effective and more engaging.”
- Ivan Crewkov
Conclusion
AI technology has the potential to usher in an era of character and relationship-based learning that could unlock something we've been missing in digital education, and Lego has shown us that infinite customization and user agency can live smoothly alongside world-class IP. But we’re only scratching the surface so far on what an AI-Native Character really looks like.
Just as the most effective human teachers create genuine connections that make students actually want to show up and engage, AI companions with faces, personalities, and the ability to remember what makes each student tick could- could- tap into that same intrinsic motivation. The question isn't whether students can develop meaningful relationships with AI characters—companion apps have already proven that emotional attachment is possible—but whether that attachment can be channeled into real educational outcomes… and whether kids can be protected from the inherent risks of such an approach.
We're fast approaching a tipping point where the technology is finally catching up to the vision. As companies like Synthesia, HeyGen and Google Veo push the boundaries of what's possible with video AI, someone will have the guts to be the first mover who creates an AI tutor that doesn't just deliver curriculum, but embodies the kind of personalized presence that makes learning feel less like work and more like spending time with someone you genuinely like and maybe even want to impress. They’ll get pushback. But the toothpaste will be out of the tube. Are we ready?
Support Edtech Insiders
Our goal is to make Edtech Insiders an enduring and sustainable community offering. If you love Edtech Insiders and want to support our work, consider becoming a paid Edtech Insiders+ subscriber for only $15/month. Paid subscriber benefits include:
Early access to all Edtech Insiders events. No more sprinting to sign up for our monthly happy hours, edtech summits, and online conferences.
Access to our exclusive WhatsApp channel to connect with edtech community leaders and discuss the latest trends and news in our space.
Edtech Insiders Upcoming Events
The Future of Equity-Based Education
How can schools support vulnerable students when equity and inclusion have become politicized? Join us for this hour-long event with two engaging segments:
A joint dialogue with Amy McGrath, VP of Educational Outreach and Student Services for Arizona State University, and Steve Levitt, American economist and co-author of the best-selling book Freakonomics
This session kicks off a 3-part webinar series leading up to EDTECH WEEK 2025, the #1 EdTech event in New York City happening October 20–22. EDTECH WEEK sells out every year! Secure your spot now and get 20% off registration with code EDTECHINSIDERS.
This edition of the Edtech Insiders Newsletter is sponsored byTuck Advisors.
Since 2017, Tuck Advisors has been a trusted name in Education M&A. Run by serial entrepreneurs with over 30 years of experience in founding, investing in, and selling companies, you deserve M&A advisors who work as hard as you do.
Not all LOIs are the same. Use our free Deal Value Calculator to estimate the relative expected value of your LOIs.
OpenAI's first comprehensive study on ChatGPT usage reveals that its 700 million users collectively process over 2.6 billion messages daily, with a substantial portion dedicated to general research, coding support, and email writing. The findings underscore ChatGPT's growing influence in both educational and professional environments.
2. Anthropic Economic Index: Understanding AI’s Effects on the Economy
Anthropic recently released a report analyzing how AI is being used across U.S. states, industries, and job types, highlighting where adoption is concentrated and how tasks are shifting from human-AI collaboration to automation. Key highlights:
It takes time for transformative technology to reach everyone, and AI is no different: right now, the use of AI is concentrated in the specific pockets of the US where knowledge work is thriving.
Some states’ usage jumps out as characteristic of the states themselves: in Colorado, use focuses disproportionately on travel and event planning, and in DC, people are especially likely to use Claude for document editing and career advice.
AI is being used to fully (or ‘directively’) automate more and more tasks over time. In just nine months, we saw directive automation jump from 27% to 39% of all conversations. For enterprise customers, that figure is 77%.
3. AI for Equity Launches AI Innovation Index to Measure School System AI Progress
AI for Equity has introduced the AI Innovation Index, a first-of-its-kind initiative to collect system-level data on AI adoption in PreK-12 schools, focusing on student AI agency, staff AI empowerment, and C-suite leadership practices. Participating school systems gain access to analytics, benchmarking, and resources to track AI innovation progress and identify exemplary practices across the country.
4. California Community Colleges and Google Launch Nation’s Largest Higher Education Systemwide AI Partnership
In a significant move to prepare students for the AI-driven workforce, Google has partnered with California Community Colleges to provide over 2 million students and faculty with free access to AI training, Google Career Certificates, and advanced tools like Gemini for Education and NotebookLM. This collaboration aligns with the system's Vision 2030 goals, aiming to enhance student success and workforce readiness across the state's 116 community colleges.
5. Navigating EdTech Leadership in 2025: Key Trends and Talent Priorities
As the EdTech sector faces tightening budgets and increased scrutiny across K–12, higher education, and workforce learning, the demand for adaptable and AI-fluent leadership is intensifying. In their latest report, NU Advisory Partners outlines critical talent priorities, including sales and go-to-market leadership, revenue operations, and strategic creativity, emphasizing the need for executives who can drive growth and innovation in a challenging landscape.
6. BOOKR Next Release: AI-Powered Reading Platform for Gen Z
Out now, BOOKR Next combines a rich digital library with interactive storytelling and AI-driven tools that make English learning more personal, engaging, and effective for teens. Backed by research showing a 20% boost in reading comprehension within 6 months, it’s setting a new standard for how pedagogy, creativity, and technology come together in education.
Abhay Gupta and Frizzle Are Saving Teachers 20 Hours a Week
We have had some amazing guests on The Edtech Insiders Podcast in the last few weeks. One of our stand-out interviews was with Abhay Gupta, the Co-Founder and CEO of Frizzle. Frizzle uses AI to grade handwritten math assignments for teachers, providing personalized feedback for every student. Teachers spend over 10 hours a week or 25% of all their work grading - Frizzle does it in minutes. If you're a math teacher or a school admin, please reach out to us at abhay@frizzle.com or try Frizzle out for yourself at www.frizzle.com.
Here’s a deep dive on our interview with Abhay, and we encourage you to give the full episode a listen for more!
Frizzle was founded to address teacher overwork, particularly the 10–20 hours weekly spent grading math worksheets. Abhay was inspired by friends leaving teaching due to burnout, and he saw an underserved problem with grading handwritten math.
“Teachers actually work like 50 to 70 hours a week…grading kind of makes up 10 to 20 hours of that a week. 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…so we kind of said it seemed like there’s a clear opportunity to leverage our tech backgrounds to really help teachers here.” - Abhay Gupta
How Frizzle Works
Teachers scan or photograph paper worksheets, upload them, and Frizzle’s AI grades them with red-ink style feedback. Students receive near-instant feedback, and teachers get both graded papers and aggregated insights on class performance.
“In about like a minute or so, you’ll get every worksheet graded with red ink feedback on the page, just like how a teacher would mark it. Not only is this helping teachers, it’s also helping students because they’re getting this individualized precise feedback that they normally just can’t get because the teachers don’t have the time.” - Abhay Gupta
Agile Learning vs. Waterfall Learning
Abhay uses his product management background to introduce “Agile Learning.” Instead of the slow feedback cycle of traditional schooling, Frizzle enables immediate iteration and personalized lessons tailored to weaknesses.
“In the school system… it follows a waterfall learning process where it’s like the teacher gives a lesson, the student gets a homework assignment, and then they get that feedback one to two weeks later…what Frizzle is pioneering is this concept of agile learning. With Frizzle we can make the cost of grading zero, and the cost of lesson planning zero, so the teacher can guide each student on their own personalized learning journey.” - Abhay Gupta
Pen-and-Paper vs. Digital Learning
Despite the push for 1:1 devices during COVID, Abhay highlights research and teacher testimony showing better outcomes with pen-and-paper learning. Frizzle bridges analog and digital by grading paper while leveraging AI for speed and personalization.
“All this research is coming out that is proving students learn way better on pen and paper…when you have a device in your classroom, students are just distracted…teachers have told us it’s way harder for a student to copy and paste something in ChatGPT when they have to hand write it. So we see this trend and we want to support it.” - Abhay Gupta
The Future of Teachers, AI, and Personalized Education
Abhay envisions AI as a “helper teacher,” freeing educators from time-consuming grading and enabling personalized instruction at scale. He stresses AI won’t replace human connection but will allow democratization of tutoring-like support for every student.
“At least for teaching, there’s nothing to be worried about with AI because teaching is a human-to-human connection…what we envision AI becoming is like a helper teacher, so it becomes very easy to manage students and help them one-on-one at scale. Students will get that individualized tutoring that most wealthy parents pay thousands of dollars for, and we want to democratize that for everyone.” - Abhay Gupta
Curious to Learn More?
You can listen to our full interview with Abhay, as well as interviews with many other edtech founders, investors, and thought leaders at The Edtech Insiders Podcast! Check it out, and as always, we’d love to hear what you think!
