Can AI Actually Teach? Inside Oboe with Nir Zicherman

Nir Zicherman is the CEO of Oboe. Oboe's mission is to teach one billion people one trillion things. In 2015, Nir co-founded Anchor, which grew to become the world's largest podcasting platform. After Spotify acquired Anchor in 2019, he led Spotify's expansion into audiobooks as VP of Audiobooks.

💡 5 Things You'll Learn in This Episode

1. Why AI should teach—not just answer questions.
2. How personalized learning paths can outperform traditional chatbots.
3. Lessons from building consumer products that reach millions of users.
4. Why curiosity and intrinsic motivation are the future of learning.
5. How Oboe is reimagining education with AI-native learning experiences.

✨ Episode Highlights

[00:02:08] From Anchor to Oboe: why Nir made the leap from podcasting to AI-powered education.
[00:06:06] Why learning is the next great consumer technology opportunity.[00:10:00] How Oboe builds adaptive learning paths instead of AI chat conversations.
[00:17:12] The biggest flaw in large language models as teachers—and how Oboe addresses it.
[00:21:13] Bringing consumer product design principles to education.
[00:26:06] The vision to teach one billion people one trillion things.
[00:31:28] Why intrinsic motivation and curiosity are the future of AI-powered learning.

😎 Stay updated with Edtech Insiders! 

Follow us on our podcast, newsletter & LinkedIn here.

🎉 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.

This season of Edtech Insiders is brought to you by Cooley LLP. Cooley is the go-to law firm for education and edtech innovators, offering industry-informed counsel across the 'pre-K to gray' spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education.

Tuck Advisors was founded by entrepreneurs who built and sold their own companies, frustrated by other m and a firms, they created the one they wished they could have hired but couldn't find. One who understands what matters to founders and whose North Star KPI is the percentage of deals closed. If you're thinking of selling your ed tech company or buying one contact Tuck advisors now.

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.

Transcript

Copy Transcript Copied

[00:00:00] Nir Zicherman: The pushback that people would say, "Well, people aren't comfortable talking into a phone and recording their voice." Well, that's not true because billions of people every day are having phone calls with each other and communicating in exactly that way. So how could you say that people are not comfortable with that?

So I think it's all about a shift in perspective. If you shift your perspective to say, "Let me look at the behavior that people are already exhibiting and figure out the inefficiencies that exist," that to me is a market opportunity. So with podcasting, it was people were already exhibiting this behavior of storytelling and communicating and talking into their phones.

They just weren't recording them and distributing them through a particular channel.

[00:00:34] Ben Kornell: 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 backchannel insights from Alex and Ben. Hope you enjoy today's pod.

[00:01:14] Alex Sarlin: Welcome to EdTech Insiders. We have a really amazing and very, very interesting guest with us today. We're talking to Nir Zicherman. He's the CEO of Oboe, that's oboe.com, like the instrument. Oboe's mission is to teach one billion people one trillion things. And Nir has an amazing background for EdTech, right?

In 2015, Nir co-founded Anchor, which grew to become the world's largest podcasting platform. After Spotify acquired Anchor in 2019 and basically became Spotify for podcasters, he led Spotify's expansion into audiobooks as the VP of audiobooks. And now he's working with Oboe, which is an incredible platform.

If you haven't yet experienced it, I recommend you do, and we'll talk about it a lot today. Nir Zicherman, welcome to EdTech Insiders. 

[00:02:00] Nir Zicherman: Thank you so much for having me. I gotta tell you, Alex, since I worked in podcasting and audiobooks for a long time, you have a fantastic podcast voice. I don't know if people have told you that.

[00:02:09] Alex Sarlin: No, I d- I've never actually been told that, and I really-- That's great to hear. Yeah. I appreciate it a lot. So thank you. Yeah. That comes from somebody with an enormous amount of podcasting experience. So let's start there because I don't think we've ever talked to somebody who came into the EdTech world from this incredible media podcasting sort of empire.

So based on your experience building Anchor, which has reached millions and millions and millions of users, and now moving into the EdTech space, how do you think about education as a scalable enterprise, as something that you can actually build into a meaningful and huge platform like a Spotify or an Anchor?

[00:02:45] Nir Zicherman: I'll tell you an interesting anecdote about the world that I came from. I was not a traditional podcaster. I didn't come from that world. I, I came from the tech world and bui-building products. I'd worked at various tech companies where we just built and scaled consumer products and tried to build products that provide a lot of value to users.

And when I started Anchor with my co-founder back in twenty fifteen, it was early twenty fifteen, podcasting was not a thing that people cared about. You know, this was right around the time that the podcast Serial, if you remember Serial, came out, and all of a sudden people knew what a podcast was, where before that, nobody really had ever...

You know, it was pretty niche format and a, a niche phenomenon. And then Serial started to make it mainstream, and people started taking this seriously. We started the company, and we went out to raise money. The first investors, the first round of funding that we raised in twenty fifteen, I can't tell you how many investors said, "No one will ever wanna record their voice."

This was twenty fifteen. This was not that long ago. But more than half of the investors that we met with would say things like, "People aren't comfortable recording their own voice. Everybody thinks they sound weird recording their..." Remember when people used to think that? Like, when was the last time you heard somebody say that whole thing?

But I remember being a kid and everybody thinking, "Oh, wow, the sound of my recorded voice is so weird." In, in today's world, that is not a thing anymore. It is incredible to think back on the fact that it's only been, what, eleven years since that mentality was so prevalent, and now it is truly something that you don't hear anymore.

It only took a few years for the podcast industry to just completely change the world to the point where it's part of a presidential campaign. It's part of public persona that major public figures have when they go on podcasts. And, and so it was just a completely different world. But at the time, people thought podcasting is a small opportunity.

Podcasting is not something that could become mainstream. Podcasting is not a true consumer consumption platform or creation platform. And we had this hypothesis that was wrong. I actually think there was a lot of common DNA between how we felt about it then and how I feel about learning now. You know, a lot of people, and I'm sure a lot of your audience, this will resonate with them People hear education and they hear EdTech and they think limited possibility and limited room for consumer reach, and I completely disagree with that.

So much so that it's baked into the mission of this company, which is we wanna teach a billion people a trillion things. You wouldn't have that mission if your view was that this was a limited opportunity. There are only eight billion people in the world, and we wanna give all of them the ability to learn powerfully using these tools.

[00:05:06] Alex Sarlin: 100%. It's interesting, when I started this podcast about four years ago, podcasting was a thing by then, as you know , but it was still something where I wasn't sure w- how I was gonna put the pieces together. And when I started looking into the tools and the suites of, of podcast directories and podcast platforms and all the things that were out there for editing services, music services, it made me really realize that it was a full industry and it was something you could step into and be supported in.

It wasn't somebody sitting in their garage editing their audio files the way it probably was 15 years ago. And I- That's right ... the, the reason I'm bringing this up, a little personal anecdote, but the reason I'm bringing it up is I think that what you've done with Anchor and what, you know, a number of the different companies who sort of entered the podcast space and made it into a full industry have done is enabled that entry, uh, you know, that access to entry for thousands and thousands and thousands of people who have become podcasters of various degrees of, of popularity, of various degrees of scale.

But the podcasting as a medium has become massive. And I'd love to hear, when you say education has that same potential, I think even those of us in EdTech don't always see it as a sort of commercially viable technology that can sort of break barriers. We've all seen Duolingo and, you know, a couple of companies sort of become household names, but not that many.

So tell us what you mean when you say, "Hey, EdTech is where podcasting was 10 years ago." 

[00:06:23] Nir Zicherman: So I think the issue is that a lot of people come at this question through the perspective of what has been done historically and what you've seen historically. But I actually think the right angle is to ask about the behavior that people are already exhibiting online, regardless of how they label it and how they think about it.

What are people actually doing online? The way I think about it is every single day, billions, truly billions with a B, people online are learning things. They learn things through Wikipedia, they learn things through YouTube, they learn things through Reddit, they learn things now through ChatGPT and Gemini and Claude.

They are using resources constantly to learn things that they are motivated to learn. We'll talk later about intrinsic versus extrinsic motivation and where that motivation comes from. But the reality is they are constantly using resources online to learn, and yet none of those platforms that I just mentioned to you were actually built to teach.

Not a single one of them, right? And so the ubiquitous platforms that exist out there, the tech products that exist that people constantly rely on to learn things online, whether it's understanding things that are happening in the news or some topic that they're interested, some niche topic they're interested in going down or some language they wanna learn, whatever it might be Very, very few experiences are actually built to teach them the things that they wanna learn.

Now, that strikes me as a massive inefficiency, right? Because it means that there's a huge market that is being underserved, and I felt the same way with podcasting. The, the view when we started Anchor was, sure, you could look at the podcasting industry just like you could look at the edtech industry, and you could say there's a cap to how many people currently wanna do this.

That's true. But you could also look at it and say, there are billions of people out there who have conversations on a daily basis, who have interesting stories to tell, who ha- who have a perspective on something that they wanna communicate to other people, who have particular interests that they wanna talk about all day long, and if you just put a microphone in front of them, they'd be willing to share it with a community that is not their friends that they're sitting in a room with.

The pushback that people would say, "Well, people aren't comfortable talking into a phone and recording their voice." Well, that's not true, because billions of people every day are having phone calls with each other and communicating in exactly that way, so the-- how could, how could you say that people are not comfortable with that?

So I think it's all about a shift in perspective. If you shift your perspective to say, "Let me look at the behavior that people are already exhibiting and figure out the inefficiencies that exist," that to me is a market opportunity. So with podcasting, it was people were already exhibiting this behavior of storytelling and communicating and talking into their phones.

They just weren't recording them and distributing them through a particular channel. In the case of EdTech and learning, it's people are already exhibiting this behavior of using all these resources online to stitch together learning paths to look, to understand topics that they're interested in understanding, but they're doing it incredibly inefficiently, right?

And so how do you get to a billion people, a trillion things? Well, you build the world's first generalized learning platform that's a consumer platform that's meant to enable all these people to learn what they want to learn more efficiently than ever before. 

[00:09:06] Alex Sarlin: You mentioned YouTube, Wikipedia, Reddit, some of the tools that, as you say, were not built for education but became incredibly important educational tools.

We've talked to the head of YouTube for Learning on this podcast, and she says at the beginning of every college year, a huge number of students go to their first class and then come back home and go to YouTube and make sense- Right ... of all the things. I mean, y- exactly. And those are all platforms that are sort of focused on peer-to-peer communication, and the learning often comes from this sort of peer-to-peer network.

AI has created this other incredible opportunity for learning, and some of the people from OpenAI and Anthropic say that the biggest behavior they see on that is learning behavior. People are trying to learn new things with AI, and that sort of is a great entry point to exactly what you're talking about with Oboe, because if people are naturally using AI to learn, which they are, it creates an incredible opportunity and a shift, a paradigm shift, where you can say, "Okay, if people are doing this naturally, what would it look like if we had a really structured, very clear, very comfortable space in which they could actually do this?"

And I think that's exactly where Oboe comes in. So for those- Right ... of you who have not actually experienced Oboe, tell us about what it is and how it works to sort of enable that kind of behavior and accelerate that kind of independent learning behavior online. 

[00:10:17] Nir Zicherman: Absolutely. So we are live at oboe.com.

We're a web app that you can access at oboe.com, O-B-O-E.com. You can enter any topic that you're interested in learning, truly any topic that you're interested in learning, because the power of AI is that we can actually offer this in a truly generalized way. And what happens is you get dropped into an experience where you are chatting with Oboe and talking to it, but rather than thinking about the product as the way that many AI products work where a user asks a question and they get an answer.

We actually like thinking about it in the inverse way, which is we are the teacher, we are the ones guiding the conversation, and you are the student reacting to the conversation with the ability to override as needed. And so we have a perspective as a platform around given your objective that you said you wanna learn, we will put together what we call learning path for you, and we will try to guide you along on that learning path by teaching you things, showing you things visually, creating artifacts that reinforce everything that you're learning.

And because AI enables us to do this, we are also fully dynamic. So what that means is that as the conversation progresses, if we determine you need more emphasis on certain things, you prefer exploring a different path than the learning path that we had set out to you. The entire system is built to constantly adapt to what it is that would give you maximum value, right?

And so the way we'd like to think about it is it sort of takes all of the benefits that you would get out of using a large language model to learn, while also solving what we think are a lot of the fundamental flaws that exist with using a large language model to learn, such as the fact that it puts too much of an onus on the learner to determine where they want to go and how they want to learn.

We want to be the ones to provide that value and then give the users the ability to override that or steer us if they feel it's necessary. 

[00:11:52] Alex Sarlin: Yes. One thing that really excites me about how Obo creates these learning paths on demand, as you're mentioning, is that it's a very interactive experience, right?

It actually creates questions. It creates a sort of a whole structure that a learner can walk through and make sense of. But it-- at every point, it's checking understanding. It's making sense of where a learner is at, of why they're trying to learn what they're trying to learn. I think it's, it's a very conversational model of how you can actually make sense of a learning path that, as you say, solves-- I think it does a very good job of solving some of the intrinsic flaws in the standard large language models, which is that they sort of just bombard you with information very passively.

It's like, "We'll give you everything, and you're gonna have to read it and make sense of it and-" That's right. "... edit what you don't like and focus on what you d--" Where it actually puts the onus, right, much more back on the learner to navigate themselves. Tell us about how you sort of design for that type of behavior.

[00:12:43] Nir Zicherman: One of the challenges of building a product like this that's a net new product, and especially one that is attempting to do something that people haven't done before, is you will inevitably get things very, very wrong. This is my second company, so I feel like I have slightly better product instinct on where to go and where to focus.

But the reality is, like, we are constantly iterating on the product and changing the product. And I'll give you a great example of how that has manifested. We have over the, let's say, year that this product has almost been in market, we, we've been out for almost a year. Over that year, we have very much shifted our perspective in terms of how prescriptive we should be as a platform around how it is that we teach you and how we guide you along this learning path.

The early versions of the product were very prescriptive because the assumption was when you rely on a teacher, for instance, it is the teacher who is doing-- overwhelmingly the teacher who is making the decisions about where to go next and how to guide you. And if you look at certain platforms out there that are already educational platforms, many of them are very prescriptive around, "If you want to get from A to Z, here's how we're gonna get you from A to Z."

And we realized that There's this paradigm shift that's happening right now in the world outside of education and outside of learning, which is that people are now becoming so comfortable with the flexibility and the adaptability of large language models that to have a platform that's overly prescriptive and restrictive around what you can do and how you can talk to it is a clear disadvantage relative to your ability to just go to an L11 and talk to it directly because of how flexible those systems are.

What we ended up doing with the product, which has proven to be a much better product as a result of this, is we are prescriptive in the sense that we have certain milestones that we want to hit. We know what the next milestone is that we want to take you to. We know where it is that we're headed beyond there.

We know what information you need to get there. But rather than thinking about it as a linear path, what we actually allow you to do-- They're, they're almost like mile markers on some journey to get to a destination, right? The path that you use as a learner to get from mile marker one to two to three is actually in your hands, and that shift is actually less of a teacher standing in front of a classroom type of mentality and more of a tutor.

If you think about the way that a tutor who's sitting down one-on-one with a student thinks about teaching, they have a plan. They have the mile markers that they want to hit, but they will adapt in real time as the user Ask questions as they want to focus on certain things, as they want to take certain detours.

And their job is not only to support the learner in going on those detours, but also finding a way to bring them back and make sure that they then hit that next milestone. And striking that balance has proven challenging. It's taken us a year of iteration on the product to, to get to the point where we're able to do that.

We're able to strike the balance of we want to be authoritative and we want to give the user these mile markers along the way, but we also want to give them the flexibility that they're now accustomed to because they use large language models every day. 

[00:15:24] Alex Sarlin: Right. That sort of tension between prescriptive structure, sort of rigid, "Okay, here you go.

Here's your every reading. Here's a reading, then a quiz, then a reading, then a video, then a quiz." Right. That structure versus a totally open-ended large language model where it just says, "What do you want to learn? What's up today?" And you start, and it's this totally free-flowing conversation. Finding a middle ground there that actually is engaging and exciting enough and sort of familiar enough for students to want to stay with it.

And I say students, it could be learners of any age, of course. They're excited, they're motivated, they want to stay with it, but also structured enough to know that they're making progress, to know that they're actually moving towards a very clear end goal. That is a tension that I think has been baked into education technology for a long time.

You know, I, I was at Coursera for a long time. We wrestled with exactly the same problems- Yeah ... a long time ago. Should we sort of create courses that felt incredibly rigid and clear because you know that you're just checking the boxes, or do we want to create opportunities for branching? And what's interesting about large language models is that opportunity for branching, that opportunity for, for flexibility, as you say, is really new.

For many years, EdTech was limited to having to be pretty rigid, having to do a lot of multiple choice questions, having to do a lot of static videos or readings because- That's right ... there was no way to know what the learner was, was picking up or what they wanted. And now we have this incredible large language models just allow that type of personalization, adaptivity in real time.

What's so interesting about, about it really navigates that tension between structured rigidity and mile markers, as you say. You, you know that you're on a path, that you're moving, you're making progress in a really clear way, which is very important for learners, but also flexibility, adaptivity. You can actually make choices within those mile markers of how you want to learn, of what you want to learn, of what you want to dig deep on.

That tension has been part of the EdTech world for a long time, and I-- hearing that you are iterating on it so quickly is really exciting to hear. My question is, you know, given this shift from what we always used to have to do, structured rigidity, and now we have this incredible flexible technology, where do you think this is all going to make EdTech much more of a consumer-friendly type of technology?

This is sort of core to your thesis, is that EdTech is something people want. They're doing with consumer platforms, but still the EdTech world has not quite broken through into the mainstream. H- do you think this flexibility is the key to making that possible? 

[00:17:44] Nir Zicherman: I think the flexibility is a massive accelerant if used in the right way.

But I think that one thing that I have seen, a trend that I've seen, is that the reliance, the over-reliance, especially for students, and I would say students in particular, but this is the case for anybody who's using LLMs to learn the way that I've been talking about learning, which is these billions of people that are using the internet to learn.

The flexibility actually, in a lot of ways, could be a detriment, and it's because it is the user who is empowered to, as I mentioned, drive the conversation forward, and it is the fundamental architecture of a large language model from a technical perspective. The fundamental architecture of a large language model is not this mile marker GPS type of analogy that I used, but rather a completely different one, which is in a large language model, if you make a left turn when you're suppo-- when it's expecting you to make-- to continue straight and you choose to make a left turn, the entire context shifts to the left, right?

And the destination that it thinks it's leading you towards is actually completely different than the one that it originally set out to teach you. That's a fundamental pedagogical flaw which assumes that a student asking a question of a tutor or a teacher should completely reorient what it is that the teacher is, is eventually trying to accomplish.

Nobody learns that way Right. The way that people learn is they have a destination in mind, and they expect that their various side quests that they're going on and their, their detours that they're taking are not going to fundamentally change the eventual outcome that they're getting to. And large language models, by design, are actually built to do the exact opposite.

As you take a detour, a large language model adapts in order to serve you towards a different destination. That, to me, is arguably one of the biggest flaws that exists today with LLMs. I think there are many flaws that could be fixed when it comes to how LLMs teach students. You touched on some of them, such as the fact that it bombards you with information, it asks you too much to be in the driver's seat and, and determine where it is that you wanna go next.

But I actually think more than any of those, the biggest issue is that it does not retain its context about its stated objective, which is where you wanna go and what it is that you wanna learn. And anybody who's had a conversation with an LLM that's lasted more than a few turns probably understands what I'm talking about.

You end up going down these paths where you're like, "I don't understand how it is that I ended up here given where it is that I started." And it makes for a very ineffective tutor because of that. 

[00:19:55] Alex Sarlin: Yes. That's incredibly well put. I think that was a- one of the most succinct and clear explanations of why standard off-the-shelf models really aren't designed for teaching.

And there, there are lots of reasons, but I've never heard that put in that way, and I totally agree. I've experienced that many times, but I've also seen learners sort of fall into these strange rabbit holes. As you say, if you're working with a tutor, especially teenagers, right? If you're working with a tutor, they will ask a question about some side issue or some tangential piece, and literally the job of the tutor is to keep the student on track toward the goal- That's right

toward the learning goal. That is maybe the main goal of the tutor, and that is exactly the opposite of how LLMs react. They're, they're sycophantic or ev- even if you take out the deep sycophancy, they're meant to react. They're built to react to what the user is telling them to do. That's right. So if you ask a tangential question, as you say, it'll turn left, it'll keep going left, it'll, it'll go in circles, and it will never sort of take you back to your original goal.

Even if you stated right at the beginning, you know, "I have a test tomorrow on physics, help me do well with that," it will just leave that behind and say, "Oh, I'll give you all this information." That's incredibly well put. You know, I, I talked to a edtech founder recently who said, "LLMs are built as productivity tools pro- for professionals."

That's what they are, right? And so productivity tools for professionals act very differently than teachers and tutors, and I think you're saying that incredibly well. One of the things that stands out about Obo, and I think it's something that edtech companies don't always think about that much, it's incredibly, incredibly fast.

When you go in there and you say, "I wanna learn about XYZ," it develops material for you and questions and responses as fast, it feels like even faster than traditional LLMs. It's unbelievably rapid, and that actually really makes a difference, experiential difference in what the software feels like. And this is something I think EdTech tools don't always think about.

Tell us how you think about that speed as a factor in the experience of learning. 

[00:21:44] Nir Zicherman: First of all, I'll say there is a very, very active tension that constantly exists there, especially in the world of LLMs, because for anybody who's worked with these models before, what they have probably seen is that the higher the quality, the slower the output, right?

That's just-- there's a direct linear relationship, in-inverse linear relationship between how quickly you are able to get something and how good the output is. And so it's always been a core product philosophy of ours to say w-we need to make this magical user experience. Speed is, is a critical component of a magical user experience, and yet we need to strike the balance of also being able to provide really high-quality output.

So when, when you talk to Obo, you enter anything into Obo, there is a pipeline that happens behind the scenes. You know, it seems as if you're basically just making a call and getting a response, but what's happening is a multi-step process that has to do everything that we've talked about in terms of distilling down the context, making sure that it retains under awareness of where it is that you are headed, making sure that it understands the steps for how to get you there, figuring out how to output the optimal next step, and doing it in a very quick way.

It's a tough thing. It's a thing that we are constantly working to make better and make a core part of the product. I think a lot of that comes from my experience building non-education products, if I'm being totally honest, right? My background is not in education, not in formal education. I, I'm extremely passionate about it and I'm a very enthusiastic, lifelong, self-driven, lifelong learner of many topics, but I don't come from the traditional education background.

I think a lot of people who work in EdTech do, and that's amazing because they bring a lot of strengths that I don't have and that members of my team who haven't worked in the space, some of them have, but members of my team who haven't worked in the space don't have that. But what we do bring is the perspective on how to build consumer products, right?

'Cause that's what I've done for my entire career. Bridging that is a really interest-- It's a really interesting marriage to bring to an industry that I think in a lot of ways is really ripe for innovation, right? And I think there's, especially now with the technology that's available, there's so much room to do things that have never been done before, and coming at it from an outsider perspective, I actually think serves as a big ad-advantage to us.

[00:23:44] Alex Sarlin: You know, it's interesting you say that because that was exactly my experience as I started looking at Obo and trying it out for myself. I said, "This is clearly designed by people who understand sort of consumer experiences." It feels the magic you mentioned. This is a, a famous sort of product metric. How fast can you get users to the magic to feel like, "Oh my God, this is incredible.

This is an incredible experience. I'm now in. I'm in. I-- the magic is there." You really do feel that almost instantaneously with Obo, and I can imagine how much work it takes behind the scenes to be able to keep all that context, to be able to navigate the LLM in all the directions, the underlying models in the directions you want, but also do it quickly and maintain quality.

That sounds- Yeah ... difficult, but at the same time, the magic is there and it, it's really interesting to feel because, you know, I think it's something EdTech companies really wrestle with. As, as a product person in EdTech, I can tell you firsthand, it is often difficult for EdTech companies to deliver that magic, that early magic- Yeah

that sort of, you know, within a minute of getting into the product, you have that wow moment and you really do feel it with Obo. And I think one of the things that's also really interesting about Obo, that magic and other things, and your background, of course, has translated into a really interesting investor.

You are actually supported by Andreessen Horowitz, one of the absolute most famous, successful venture capital firms in the world. Can you tell us a little bit about that experience? Because I think that also is a great example of your sort of bridging between education and consumer technology, and I think they recognize how strong that combination is.

[00:25:09] Nir Zicherman: Yeah. I think the team at Andreessen had been looking... My understanding is that they had been looking to make an investment in education for a while, especially with the advent of artificial intelligence and all the stuff that's been happening over the past few years. They've been looking for a company to do it.

And I, I think one of the, from conversations I've had with them, it seems like One of the biggest issues with the companies that have pitched them, I think, is the level of ambition, right? When you're talking to a venture capitalist, they're-- obviously they want... They're thinking at scales that are astronomical in a lot of cases 'cause they need to return their funds, and that's just the way the math of the business works.

A lot of education companies that go out and try to raise venture capital, I think are actually not venture backable businesses because they could ex- be extremely healthy businesses, but they're not Looking for the type of escape velocity approach that a massively scaled consumer product has. I think that's one piece of it.

The other thing is, it's a really interesting time right now where venture money is actually disproportionately flowing to B2B businesses and enterprise businesses. It's not flowing to consumer businesses. There are just not a lot of founders that are trying to build very large scale global, truly consumer businesses.

And I think that's one of the things that's probably unique about us is that we are trying to tackle what, from my perspective, is a global problem, and it's a massive opportunity, and it's a much bigger opportunity than I think a lot of people have given credit for. And I think that that level of ambition is probably what resonated with that team and why they were excited to partner with us.

[00:26:32] Ben Kornell: Yeah. 

[00:26:32] Alex Sarlin: It, it makes a lot of sense. And I think that is a great segue to your sort of core mission, which you-- we stated in the, in the intro and you've mentioned, but I wanna dig into it here because I think it's something that, as you say, is not always the core ambition of education technology companies.

It's not always how the industry sees itself. You, you have this mission of to teach one billion people one trillion things. That's a, you know, significant percentage of the world population. When you're talking a billion people, you're talking about consumer platforms that are massively scaled that, that reach a billion people.

So what inspired that vision to, to think at scale from day one, and how does it influence the way you're actually building the platform? How are you thinking about unlocking that latent demand that you're seeing from people to learn, to figure out how to navigate the world, to learn new things? You're seeing it in YouTube, you're seeing it in Wikipedia and Reddit and all over the place.

How are you looking to channel that demand into Obo? 

[00:27:24] Nir Zicherman: Earlier, we mentioned this concept of the inefficiency that exists when people are trying to learn online, right? This notion of teaching a billion people a trillion things doesn't sound crazy when you think about it through the lens of there are billions of people exhibiting the behavior that we're trying to capture online every single day.

And here's what they have to do in order to succeed in their desire to learn whatever topic it is. They have to manually go out and stitch together resources online and create the very learning path that Obo tries to create for you. They have to do it manually themselves by finding disparate resources on the internet.

And not only that, but up until the moment that LLMs became a viable thing to plug into that, they had to also rely on the content that was created by other human beings. And so if you were a person who was interested in a niche topic, or you were a person who was interested in a not niche topic, but you wanted to apply it in a personal way, you couldn't do that, right?

You would basically just have to go out and find the resources that existed on YouTube and Wikipedia and Reddit and whatever else, and you would rely on the output of other people in order to consume content that you wanted to consume. When I look at how many people every day exhibit that type of behavior, all I see is inefficiency.

And so I think that through that lens, the idea of reaching a scale of hundreds of millions or billions of people using a product It's not like we're trying to invent a new behavior. This is a behavior that people already exhibit every day. They're just doing it very inefficiently. And when I think of consumer products that have succeeded at the scale that we're attempting to succeed at, it's because they did exactly what I just said, right?

They identified a behavior that already existed, and they were able to find the reasons why it was highly inefficient and tackle each one of those. That was our core philosophy at Anchor, too. Podcasting in a lot of ways, it, it was an industry that had already been around for 20 years. It was just relatively small.

And there are two ways to look at a small industry. One is it's small because it's small. It's small because it's limited. The other way to look at it is it's small because there's latent demand there, but it's not being served because there are massive hurdles that prevent growth And so our entire philosophy with that company from day one was, let's just find every hurdle and just knock it down.

Any reason why a person might struggle to create and distribute and monetize their spoken audio content, let's just get rid of it. In a lot of cases, this actually, it's, it's very relevant to the EdTech community. In a lot of cases, those hurdles did not exist for technical reasons, they did not exist for business reasons.

They existed for legacy reasons, right? They existed because it's just the way that things had been done for a long time. And I think there are huge parallels between the two industries, podcasting and education, for that reason, right? 'Cause in, in the world of education, I have to imagine so many EdTech companies right now are asking, "How do we leverage AI?

How do we incorporate it into our product? How do we build all this magical stuff that you and I are talking about?" And I actually think the biggest flaw that a lot of people thinking about this make is that they think about it through the lens of what has already been done and how they can incrementally change the things that have been done by just packaging some AI thing on top of it.

We have a lot of disadvantages being outsiders and being a small company that's new to this space, and I won't list out what those are. But I actually think we also have one huge advantage, which is that we're not bogged down by all of the industry legacy stuff that already is there. That was the case with Hanger, and I think that's also the case with Obo.

[00:30:27] Alex Sarlin: That's a really fascinating claim, and I, I agree. I think there's a huge amount of sort of baggage in, in lots of different industries, but definitely in the EdTech industry of what works, what distributes, what sells to schools, that's a big part of it, what sells to consumers. People have all these assumptions and these sort of full systems baked in about how to do this.

And one of them that I think is really-- that's really deep, and I'd love to hear you talk about it, is this motivation piece. I think that because a lot of education technology companies are designed around formal education, and formal education is sort of by nature extrinsically motivated, right?

Students have to get grades. Students have to get to the next level of, have to be promoted, have to get into college. A lot of the most successful EdTech, consumer EdTech companies tend to be sort of extrinsically motivated, right? You are really m-moving in a different direction than that. You're saying, "Well, there's actually a huge intrinsic motivation for learners all over the world.

They're trying to learn new things. They're trying to stitch together learning paths and piece together different videos and newsletters and podcasts to make sense of the things they're trying to learn." This is a latent and intrinsic demand that's massive. There's intrinsic motivation. Talk to us a little bit about how you think about the motivation of the Obo learners.

Do they come to you with a very specific goal? Are you helping them form that goal? Is it a, "I have a test tomorrow," or is it, "I wanna learn this for career"? How do you think about motivation? 

[00:31:47] Nir Zicherman: So two thoughts here. One is, I like asking people of the things that they've learned in their lives, right, the things that stood out to them.

If you think about the things that you were extrinsically motivated to learn because there was some outside force telling you, "You need to pass this test. You need to get this certification. You need to graduate," whatever it might be, and you compare it to the things that you were just motivated to learn because of a love of learning that topic, and you were just interested in that topic.

Compare how well you remember the things that you learned extrinsically versus the things that you learned intrinsically. And I'm willing to bet that there is a delta there of 10X if not 100X factor where the things that you remember from 20 years ago that you learned or in your adult life, right?

Maybe not necessarily when you were in school, but just things that you learned, I'm willing to bet that it's the intrinsic things that stuck with you. And it's because most of the behavior that people exhibit when they learn is due to intrinsic drive. When they-- you go out and you go down a Wikipedia rabbit hole, you might land on Wikipedia because something extrinsic triggered you to come to the platform.

But the reason why you stuck around and started clicking on all these links and you lost an hour of your time reading all of these pages that you weren't expecting to, is because you suddenly were intrinsically motivated to follow paths of curiosity. I think many learning products do not focus on the right half of that.

If you want to motivate a billion people You need to tap into what actually motivates a billion people. And what actually motivates a billion people are the things that they're intrinsically motivated to learn because they're curious about them, right? And they want to spend time doing them. I think it would be too difficult for us to reach the level of ambition that I've been painting for you if we were just trying to tap into extrinsic motivation.

A lot of the ways that we think about the platform is that the reason why somebody may come to the platform is for extrinsic reasons, but the reason why we want to keep them on the platform is for intrinsic reasons, right? Because we want to give them that magical Wikipedia rabbit hole or YouTube rabbit hole type of experience.

Even though in many cases, especially when those platforms first emerged, the reason why you went to Wikipedia or the reason why you went to YouTube was for a specific thing that you just needed to get done. But the reason why you stuck around with those platforms and then happened to come back later when you were bored was because of the fact that you have 100 other things in your mind that you're also intrinsically motivated to learn, and you care about those a lot more.

[00:33:54] Alex Sarlin: Fascinating to hear. I definitely agree. You know, self-determination theory, the sort of theory, the sort of the main theory behind intrinsic and extrinsic motivation, it has always been incredibly interesting to me. And one of the things, because of exactly the tensions you name, right? Because in formal and informal education, you have this huge blur of different types of motivation.

And even within an extrinsic model, so within a, you know, a formal education, maybe even a class that people are taking because it's a requirement, there are these sort of pockets of opportunity for intrinsic motivation and being like, "You get to choose the topic of your final paper. You get to choose the..."

And people only remember that. That's the only thing they remember from that class. One of the types of motivation that I always think is under-discussed is in that same theory, there's this concept of sort of integrated motivation, which is sort of in between. It's this idea of when somebody begins to identify their own personality with what they're learning, it, it's not fully extrinsic, it's not fully intrinsic, but it's a way to get from extrinsic to intrinsic, where you say- Mm-hmm

"I care about this. I think of myself as somebody who loves X. I think of some-- myself as somebody who wants to be an X." And as soon as you begin that identity formation around a topic, that is sort of the path to intrinsic motivation. And I'm curious if you see any behaviors that make sense, that connect to that when you look at your Obo user base, people who are like, "I may have come here for a certification, but what I really care about is thinking of myself as a fill in the blank, and that's where my learning is taking me."

[00:35:22] Nir Zicherman: That we see all the time. If, if you look-- We talk to users, for example, where we sit down and we look at what it is that they do on the platform and, and how they've been using the platform. And we often discover situations where somebody comes in for exactly what you're talking about. You know, I came in because I'm, I'm studying for a real estate license exam, or I came in because I'm studying-- I w- I wanna ace my law school test.

And then you look at their history of different chats that they've started with Obo, and it starts with that, and then it ve-very slowly starts to turn into things that maybe are s- related, but not exactly directly related to the certification. And before you know it, they're asking about, like, the history of some pop culture phenomenon and going off on all these interesting tangents.

And it just speaks to the fact that that is how people wanna spend their time, right? I think for your audience who are working in edtech, especially more traditional edtech companies, one thing I'd always be curious to ask them is how they think about this. Because the way that I feel a lot of edtech products are built, they try to very quickly identify what the extrinsic motivator is that brings you there, right?

And they build the entire experience around that. So if you go through an onboarding experience where, you know, you're asking the user what test they're studying for and what grade they're in and what experience they have, that's all great, but that's only extrinsic information. You're not actually identifying anything about what's intrinsically motivating them to learn.

And this is an example of something that now with the world of AI, you can so much more easily extract that information by having a-- letting the u-user in natural language explain that to you, and then you can extract from that ways to build intrinsic motivators into the experience to really make it a ten, hundred X more powerful experience than it was before.

[00:36:54] Alex Sarlin: Yes. It's really, really interesting. You know, we, we wrestled with this all the time at Coursera for exactly the reasons you're saying, which was it was so interesting. You know, the original set of Coursera courses were very-- they were Egyptology or history of the world. They were things that were really fascinating, but didn't often connect to any sort of direct tangential outcome for learners.

And then over time, the platform shifted more and more towards those extrinsic motivators, towards technology and business classes and data science. And-- But the dream was always to put these things together, just as you're saying, to-- for people maybe to come for a data science class and then find that Egyptology class and find that equine nutrition was a big hit early on, and just find these incredible- Interesting.

things that people really, really loved. I think that LLMs really help unlock that possibility because they allow people, as you say, to come in for one thing, but then to be-- for it to be conversational, for it to be back and forth, for the system itself to sort of help users unearth their intrinsic motivation, what they care about.

And as you mentioned earlier, that's what great tutors do as well, right? Tutors come in and say, "Yeah, maybe you're trying to learn math, but what you love is ballet. Well, maybe what you're trying to learn is history, but what you love is video games, or you love role-playing games." Connecting the intrinsic and extrinsic is something that humans can do very well, but that systems could never do in the past, and now they can.

And I'm curious how you're thinking about that. As you get to your billion users, how can the system itself sort of merge these motivations in really interesting ways? 

[00:38:19] Nir Zicherman: So I think LLMs have-- What we call LLMs are actually multiple benefits, I guess, that are suddenly unlocked through the same technology.

The technology happens to enable a bunch of different things and also enable a bunch of disadvantages like we talked about, right, for the learning use case. But in terms of the advantages, I think it's important to compartmentalize these different advantages and think about them in, in different ways.

One of them is the ease of accessing information on the web. It's a more efficient way to access information on the web. These models are basically trained to very, very efficiently serve information from corners on the web that went into their training data. That's great. That's one motivation. So now all of a sudden, all of the niche topics that sit in corners of the web you wouldn't have found before are now accessible to the user.

But I think what you're touching on is actually a totally different benefit, which is the ability for the first time ever to extract from natural language semantic meaning in a way that previously was not possible. And what I mean by that is If you think about any product that was built prior to five years ago, any tech product, it is the system that determines what the options are that the user can choose to give signal to the system.

The user then has to conform to that language, and they have to learn how to speak that language. And today, the exact opposite is true, which is it for the first time ever, it is possible for a person in their own words to explain something, and then the system figures out how to translate it into its language, right?

The user does not have to learn a new language. They speak whatever language they're comfortable speaking in terms of natural language. And I think now all of a sudden, that enables this whole intrinsic thing in a way that previously was not possible. Because how could you have captured intrinsic motivation previously if the only way that a person understands their intrinsic motivation is through a language that they speak with words that they speak, and which don't perfectly conform to the six options that you have on your screen during your onboarding flow?

The ability to then put it in the hands of the user and say, "Use your own words. Be as, as succinct or as verbose as you wanna be about this. Tell us as much context as you want. Tell us as many stories as you want." It doesn't matter because what we can now do is take all of that fluid, amorphous natural language and convert it into something that's actually actionable for the platform.

And I think that's a really interesting inversion that actually empowers users in a way that previously was not possible, right? Because as a user, you don't have to adapt your desires and your motivation or whatever it is to something the platform expects of you. It's actually now the exact opposite.

The onus is on the platform to determine how to take your natural language and turn it into something actionable. 

[00:40:43] Alex Sarlin: Yes And the platform could also solicit some of that information and say, "What is your favorite thing to do when you have a-" Exactly. So yeah, so it's incredibly powerful because-- And that becomes context that can become meaningful, it can become actually part of the learning experience in the current moment and down the line, right?

It could be, "Hey, I learned about you because you just told me in natural language that you really care about soccer and the World Cup." Yep. A month from now, I'm gonna bring up, "Hey, there's a really interesting-- The Euro Championship is on. Why don't we talk about that?" It, it's- Exactly ... it's incredibly interesting how that works, and I think, you know, people are really trying to explore.

All of EdTech is trying to figure out how to make sense of that. But I think what's interesting about Obo is you're AI native, you're consumer native, that's sort of baked in from day one. You're like, "How do we make sure that we're making sense of the users, of what they need and what they want and what motivates them- That's right

immediately, rather than sort of coming from this legacy system of having to make the options internally and then have users choose and type in what grade they're in or type it, you know, choose from four things instead of saying, "Tell me anything, literally anything." It's incredibly exciting. So last major question for you, and then I'm gonna just-- we can close up with our questions we love to ask founders.

But curiosity, you have mentioned curiosity in this, and this is a word that I think has come up a lot recently, and for very good reasons. Curiosity is sort of at the heart of intrinsic learning, right? People get curious about the world, especially children get curious about the world, and that's some-sometimes it's sort of beaten out of us as, as adults.

We, we sort of learn to say, "Okay, what if... I may be curious about this, but I have work to do. I may be curious about this- Yep ... but I have chores to do." I think AI has incredible power to reignite curiosity, and I know you share that. Tell our listeners about what role curiosity plays in the Obo story. 

[00:42:24] Nir Zicherman: I actually think the way I like to paint this is that they're basically-- There's a very positive view you can take on the future right now, and there's a very negative view, right?

And that we are at the crossroads of two very, very different potential futures. One is the negative future that we keep reading about where students are not learning, they're not studying, they're using AI to cheat more and more, they're using AI to write their papers. All of the motivation to do things are basically sucked away from them.

I do not believe that. And the reason I don't believe that is because of this extrinsic and intrinsic thing that we've been talking about. When you are extrinsically motivated to do something, you find shortcuts, and the shortcuts now are available to you in a way that has never been possible before.

When you are intrinsically motivated to do something, you don't find shortcuts because you enjoy doing the thing for the sake of the thing. You wanna learn for the sake of learning I think we run the risk right now of surrendering to that first possible future. And I think more than anything, the main reason why I'm excited to build this product and to build this company is because I wanna bring that second future into existence, and I want people to realize that actually learning can be motivating and fun for the sake of, of learning when you tap into the types of things that people are intrinsically motivated to learn.

I'm a parent, so I have two young kids, and they're very curious, and they ask questions all the time about the world and about what's happening in the world. And my wife and I have this philosophy of we don't want them to lose that, especially in this world where it's becoming easier and easier to not be curious and to, to not ask questions.

If they ask us a question, no matter what we're doing, we're gonna stop and we're gonna answer their question. And I think adults need that too , you know? I think everybody needs that. I think the ability to pause and follow the things that you're curious about and that you wanna learn and that you see in the world, that's what makes us human.

If we lose that, I don't really know what any of this is for, if I'm being honest. 

[00:44:08] Alex Sarlin: I love that. I, I also am a parent of two young kids. One asked me this morning, "Why are firemen's hats red?" And I have no idea, but I am making sure that that question does not leave, and that we're gonna be looking that up tonight and making sense of it because I totally agree.

The curiosity and that ability to just figure out and explore the world is incredibly powerful. I agree with you down to the core of there's a lot of concern right now about that first possible future, a world in which students are all offloading all their work, they're learning less and less, they're more-- less and less interested in any kind of learning because everything goes-- their thinking happens entirely through AI-mediated platforms.

That's right. And but I agree with you. Th-that's really a response to extrinsic motivation. And schools and universities have become factories of extrinsic motivation, right? People look at a lot of education as very transactional. And in a transactional extrinsic world, yes, you're gonna take shortcuts especially- That's right

if they're readily available, incredibly powerful, have the entire internet plus as a source material. But if we can tap into that intrinsic motivation, those curiosity, I think we have an incredibly bright future for edtech and for learning, just for learning as a concept. I think there could be a, a heyday for learning.

I really appreciate your time. This is fascinating. Nir Zicherman is the CEO and founder of Oboe, O-B-O-E, dot com. Oboe's mission is to teach one billion people one trillion things. They're backed by Andreessen Horowitz, among others, and are doing real interesting work. And I mean it, the magic is there. If you haven't gone onto Oboe and tried a trial account and actually seen how it works, it may not be clear th-the magic is there.

The speed is part of it, the comprehensiveness is part of it, the interactivity. Thank you so much for being here with us on EdTech 

[00:45:50] Nir Zicherman: In- Thank you so much for having me. I appreciate it. 

[00:45:52] Ben Kornell: 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.