Shaan Puri presents a countdown of ten AI startup ideas, from eliminating wait times with AI sales agents to warehouse robotics and AI tutors. His number one pick is the “self-doing to-do list” — AI agents that take a goal, generate a plan, and execute it autonomously, which he believes is the defining opportunity of the intelligence wave.
Speakers: Shaan Puri (host, solo episode)
Introduction — Why This List [00:00:00]
Shaan: All right, I’m setting my timer. I’m going to give myself 31 minutes to give you 10 specific AI business ideas.
I’m doing this because I see a lot of stuff on YouTube that’s like, “Hey, AI is the next big thing,” and then you’re like, “Okay, what? How? For me, what should I do with it?” And there’s no answers. Or “AI is gonna kill us all” — really? How? “We don’t know.” I just don’t like the general enthusiasm without the specifics. So this is all about the specifics.
I think I’m well qualified to do this because my whole life I’ve been an entrepreneur and I’ve been an idea guy. I ran an idea lab where I was basically funded to just come up with business ideas and then build them for six years. I then have a podcast — this podcast is all about business ideas. I’ve also invested in the last year maybe $2 million into AI companies. So I’ve got about a dozen companies that I said yes to, plus probably another hundred or so that we passed on. I’ve seen a bunch in this AI space. I know what the best entrepreneurs in Silicon Valley are doing — that’s where I live. So I wanted to share with you my list from number 10 to number one, number one being the best, the biggest opportunity that I see in the space. If I wasn’t just chilling out rich and happy and being a dad, I would go and jump in and do these right away. Instead, I’m going to invest from the side. I hope somebody takes an idea from this list and just goes and does it.
A Quick “I Told You So” [00:01:30]
Shaan: Before I tell you the ideas, a quick little history lesson, a quick little trip down memory lane, a little “I told you so.”
Three years ago on this exact podcast I did an episode called “Is GPT the Next Big Thing?” I think it was like episode 94 — we’re on like 500 now. This was back in July of 2020. COVID had just hit, people were freaking out, but me and Sam got on the podcast and we said, “Is GPT-3 the next big thing?” We had gotten access to this tool. There was no ChatGPT at the time. You had to get a favor from somebody to even be able to try it. And we used it to write rap lyrics, which of course being idiots that we are, we thought was the best thing we could do with artificial intelligence.
We thought it was amazing. We were like, “Wow, this is really good.” The episode is “Is GPT the Next Big Thing?” — we were basically saying, “Yeah, it will be.”
Now that’s 2020. Fast forward today: OpenAI is worth $90 billion. They’re doing a billion dollars a year. ChatGPT is the fastest growing product in history — hit 100 million users. So I’m not saying that we called it. I’m not saying that we’re business geniuses. We kind of did and we kind of are.
If you had listened to me then, you probably could have made a bunch of money. Listen to me now, because today I’m going to share with you 10 specific AI business ideas ranging from simple doable things for somebody who’s non-technical all the way to moonshot really big ideas that I think are game-changing. I’m going to read you the list from number 10 to number one. They include things from robotics and therapy and celebrities and education to porn — I got everything. And the last one, the very last one, I think is the biggest idea that exists today. The biggest opportunity that exists today for any entrepreneur.
When you think about Zuckerberg stumbling into the social networking craze, or when Steve Jobs created the iPhone that changed everything — there are these “change everything” moments. I was a kid when the internet came out. I’m 35 years old now, so I remember being on the internet as like a 12 or 13-year-old, using very basic dial-up connections and BBS forums and stuff like that. It was clear to me that the internet was a thing. It’s just I was a kid, so I didn’t think about how I could benefit — I just wanted to use it. I wasn’t an entrepreneur. I was a teenager. I couldn’t even spell entrepreneur. Still can’t.
When I graduated from college — I graduated 2010 — the iPhone had come out the year before. I think the App Store had come out. But I didn’t have the light bulb on. I didn’t really understand that there are few times — what’s that Warren Buffett quote? Something like, “There are moments every once in a while where the skies get cloudy and it begins to rain opportunity, and when it does, let us not run out with spoons but with bathtubs” — to catch the opportunity.
I didn’t really realize when mobile was the big opportunity, and it was right in front of my nose. We were working on things and we had a mobile developer — one guy — and we would try to make a mobile version of our website. But we didn’t realize: yo, mobile’s changing everything. All of a sudden, this thing’s got a GPS — that means now Uber is possible, Google Maps is possible. This has a camera, so that means Snapchat and Instagram are now possible. I’m carrying this with me everywhere, so all of a sudden WhatsApp is possible. This thing has an accelerometer inside, that means MyFitnessPal is possible. The technology unlocks the opportunity.
Idea #10 — Eliminate the Wait [00:05:30]
Shaan: So now to my specific list of 10 ideas that I think anybody could do. And here’s my disclaimer: these specific ideas don’t matter. I know I’m hyping them up a little bit, but it’s not the ideas that matter. Naval has this great quote where he says, “You don’t read books for the information. You read books because books spark ideas in your head.” Reading a great book is like lighting a match or starting a bonfire in your brain. That’s my goal. I want to start a little bonfire in your brain. I want these ideas to help you get to the real ideas. I don’t think that these are necessarily going to be the ones, but I hope that they really stimulate you to think.
Okay, let’s jump in. Number 10: Eliminate the wait.
We think of the internet as this fast place, a place where you can just click and you instantly get what you want. And that’s true. If you just compare to normal business — I used to have to drive, go to my car, drive to a store, park, get out, walk in if I wanted to buy something. Now with the internet, I don’t have to do any of that. Website, click, boom — it’ll be here in two days. That’s kind of amazing. The internet has killed waiting in a bunch of different ways, and that’s generated billions and billions of dollars of value. But all waiting is not gone. There’s still some waiting left.
For example, one of my companies is called Shepherd. I’m a minority owner. I bought a minority stake earlier this year. They do a very simple thing: they find you employees overseas that will cost you 80% less than if you’re hiring in the US. They’re like recruiters that will find the best talent in the Philippines or Latin America that just solves whatever problem you have. Great. Now when somebody comes to this website — this business works, it makes millions of dollars a year — but it also loses, I believe, millions of dollars a year because of one very simple problem. When you click “start hiring here,” it’s going to tell you to book a call. It says, “Great, tell us some information.” Once you fill this in, it’s going to tell you to book a call, and that call might be 2 days, 3 days, 4 days from now. By then you’re cooled off. Any salesperson knows that you need to strike while the iron’s hot. You’re going to have the highest conversion rate in that moment.
So let me show you what an AI company is doing that I think is very smart. There’s some company called Same Day — I’m not involved in this at all, I just think it’s a very cool idea. They’re like, “Hey, what if we just had a phone agent, an AI phone agent, that could reply to anybody who calls you right away?”
Listen to this: “How can I help you today?” “Yeah, I got a call from my wife a minute ago. She said there’s a bunch of ants coming through our kitchen window. How soon can you guys come out?” “That’s the worst. Luckily, I can probably have someone out as soon as tomorrow to take care of those ants. Can I get your address to confirm?” “Yeah, it’s… but can you transfer me to a person?” “Yeah, I can transfer you to a person, but you’ll be on hold for about 9 minutes. Now I can answer any questions and get you scheduled in under two. How does that sound, Aaron?” “It sounds good, but what’s it going to cost?” “Oh, of course, I just need to know…”
You get the idea. I think it’s funny they made him have an accent. But instead of “I’m interested” and then a three-day gap, now we get on a Zoom call and you’re trying to remember the context and then we fumble and bumble — how about right when I say I’m interested, an AI sales agent is going to call me or create a voice call right here on the browser and ask the questions needed to qualify me as a potential customer, as well as answer the questions that I have, and even give me a sales pitch?
You might look at that and be like, “Oh, it’s a little slow the way it talks,” or “It’s not as good as a human.” I would tell you two things. Number one, the conversion rate in the moment versus the drop-off of people bouncing because they don’t want to book a call, or not showing up to the call — the show-up rate might be 65% — you get these huge drop-offs for every additional step in the funnel. You want to just remove steps in any funnel. I believe that an AI sales agent will be higher converting than doing it the slower human way.
The second thing: this is V1. You remember the V1 of anything — the brick cell phones, or even V1 of the iPhone compared to what we have now on iPhone 15? Just wait two years. This is going to be incredible. And it’s already good enough. How many businesses don’t even have a phone number or don’t answer the phone when you call them? Every pest control service or lawn care service or self-storage facility — you name it. You can go and become the AI phone sales guy for every business on the internet.
That’s one opportunity that’s there for the taking. Forget this whole idea of “call me, I’ll call you back” or “book a call in a few days and we’ll talk.” No, it’s going to be: we’ll talk right now with my highly trained sales agent who always sticks to the script, is always polite, never gets frustrated, never gets sick, never takes a day off. That’s what’s going to happen here. That’s where this is going.
Idea #9 — Therapy for Everybody [00:10:30]
Shaan: Idea number nine: therapy for everybody.
Therapy used to be pretty taboo, and every year it’s becoming less and less so. Every year more and more people are going into therapy. There are multiple billion-dollar therapy startups that just connect you with a therapist. However, that’s still only a fraction of the opportunity. How do you make this 100 times more accessible?
I think most people could probably benefit from having somebody to talk to — somebody who’s there for them, supportive, asks good questions, gives good advice. That seems like something that’s going to help every couple, every individual person, every executive. The biggest barrier to this right now is cost and the friction or privacy that’s involved.
It’s almost like — remember how Hims came out? They were like, “You know what? Erectile dysfunction is a big deal. It’s a big problem. But people don’t want to go to the doctor, admit this to another person, and then have to go to a pharmacy and say, ‘Hey, can I get my pills please?’” Instead, what if you could just get diagnosed online, get the prescription really quickly from the comfort of your own home without going anywhere, and then they’ll deliver it to your doorstep in very discreet packaging? Hims and Ro did this and built billion-dollar companies just on that one way of increasing access.
I think for therapy, you’ve got to do two things: decrease the cost and increase the privacy. Solution: put an AI therapist in everyone’s pocket. You train it on 100 million hours of therapy transcripts and conversations that exist or could exist, and you provide the service to everybody — even people who can’t afford a $100 an hour therapist or $200 an hour therapist. You drop the cost by 100x. How do I get this for $10 a month or $5 a month? That’s the big opportunity. And in doing so, you would 100x the number of people that get this benefit. It’s a win-win.
I don’t know the exact specifics. It’s going to take some time to train this to be good. There may be some things like — you know the difference between creating drugs or a supplement? You have to either do clinical trials and get FDA approval, or you just create a vitamin and you can sell that online tomorrow. It might have to be framed less medically and more like a life coach. I’m not sure. But the idea of providing therapy for everybody is a big idea.
Idea #8 — Robots That Automate Warehousing [00:13:15]
Shaan: Number eight: robots that automate warehousing.
For my e-commerce business, I was running a warehouse for I don’t know, a year, year and a half. We had a 10,000 or 15,000 square foot facility here in California and we had 10 to 20 people there. Absolute pain in the ass for everybody. Nobody liked it. They didn’t like working there. We didn’t like having people there. We didn’t like running the thing. It was slow, it was expensive, it was bad in pretty much every single way.
Then you look at Amazon. Amazon has invested billions of dollars — I’ve read anything from $10 billion all the way up to $100 billion as an estimate — in R&D around their warehouse automation. Back in the mid-2000s, they bought a company for almost a billion dollars — Kiva Systems — and those are those little sleds. You can see these insane videos online. The sled goes into the warehouse, finds the set of boxes that somebody ordered something from, picks it up, drives back. There are hundreds of these going at once. They’re all part of one big brain, so they know how to never bump into each other. They drive all the way up to a human who’s sitting in a chair who just pulls the item out of the box, puts it in the package. That’s the Amazon product.
This is a perfect example of my “export framework.” My export business framework is basically a way to generate business ideas: you look at any big company and you see what did they spend millions of dollars building a homegrown solution for — something that works for them and made their life better — and then can you export that idea as a product that any company could use without having to spend the money on R&D?
Tons of examples of this. A simple one: LaunchDarkly, a billion-dollar company now. They were working at Facebook. Facebook had built internally — spent a bunch of engineering resources building — a way to launch new features under feature flags. Meaning you launch a feature, it’s in the app, but you can turn it on for 1% of the population. If it’s bug-free, then you can turn it on for 25% of the population. Or if it starts to have a bug, you can turn it off quickly, remotely from your server. The idea of feature flags — they productized it, took it out there, and said, “Hey, the thing that Facebook uses for their app, you should use in your app.” And of course it works.
I think somebody’s going to do that here with Amazon’s warehouse technology. I’ll just give you a crazy stat. This stat is the business plan for this business. I talk about this idea of “one-chart businesses,” “one-stat businesses.” It’s a single stat that basically encapsulates the entire opportunity: today, 2% of all warehouses use robotics. 2% of all fulfillment warehouses are using robots today. That number is going very close to 100%.
So just the shift — what’s going to take it from 2% to 100%? You could just sit down and brainstorm. Is it new robotics technology? Is it better sales and development process? Is it a consulting practice? What are the different ways you could take that number from 2 to 100? I think that’s just one big opportunity.
Idea #7 — McKinsey for AI [00:17:00]
Shaan: Number seven, similar to what I just said: consulting — McKinsey for AI.
Technologies come, but they don’t get everywhere all at once, right? It’s not evenly distributed. There’s still like four million people with dial-up internet using AOL to get online right now in the United States. It takes time for these technologies to propagate through the community, through the population.
AI does all this amazing stuff. It can help businesses become more efficient, more smart, more intelligent, serve their customers better. But it’s not just going to appear overnight in every company. No way. I think somebody can build a killer combo of conferences, content, and consulting — the three C’s — package those together and build something.
Here’s what I would do if I’m working at McKinsey. I’m one of these smart people who gets a job at McKinsey, and you have two choices. You’re either going to grind the McKinsey ladder for the next 5 to 10 years with the end goal of maybe becoming a partner someday and making a million bucks a year. Or door number two: quit your job tomorrow and launch an AI-specific consulting practice that is going to identify one type of customer — maybe it’s mid-market industrial companies, or it’s law firms, or dental practices, or whatever — and identify AI-market fit. Identify one AI tool or process that would help one type of customer and start there. Start consulting there. You can build a multi-million-dollar service business from there and stack more and more.
Basically, in the same 10 years, instead of just grinding it out at McKinsey, spend five years doing this on AI, building the AI consulting company, and sell it back to McKinsey for a 100x or 1,000x payout. I believe that there’s going to be billion-dollar consulting companies that are specifically just about AI. The big consulting companies — Deloitte, PwC, McKinsey — they’re going to need to buy. They’re going to acquire practices that do this.
If you have the chops to do it — conferences, content, and consulting, a combo of those three would work. Specifically around conferences, what I don’t mean is what I think everybody wants to do: create “The AI Conference.” That’s cool. Somebody’s going to do that. But that’s also very crowded. Instead, I think the easier opportunity is to create the “AI for X” conference. “AI for Healthcare Conference” — hey, you have a healthcare company? We have a conference that’s specifically about marrying the best AI companies and startups and experts to healthcare companies like you, and seeing how you guys are going to be using AI in the next three years in order to build a better company.
You could do that with every niche, every industry. You could do this in healthcare, in farming, in any vertical industry. You could create conferences around that. I think that’s separately just a good idea.
Idea #6 — The New Cameo [00:20:30]
Shaan: Next one: Cameo.
Cameo was this app that got really popular because it let you buy shoutouts and greetings from celebrities. Great. But if we just step back for a second — influencers and celebrities make money off their name, their face, their voice. Today, if they want to do that, they have to cut these deals with companies. They have to agree: “Hey, you’re going to have to fly out to Tucson, and we’re going to film this commercial, and you’re going to be here on set all day, and you have to read these lines.” It’s a pain in the ass. That’s why you have to charge a bunch of money for it.
Now that AI is here, we have something called deepfake technology, which basically means you can make a fake video of anyone’s face. You’ve probably seen the deepfake Tom Cruise, or you’ve seen the deepfake music video where the guy’s face was shifting from one celebrity to the next. Indiana Jones, I think, used deepfake tech to have the young Indiana versus the old Harrison Ford. Deepfakes are getting really, really good.
This is an opportunity. Now here’s where everyone gets it wrong. Entrepreneurs think — I talked to a lot of startups who said this, they’re pitching me for investment — “We’re going to make this deepfake tech, it’s going to be great.” This is not a tech problem right now. It’s actually a rights problem.
The move here is not to go all in on product, but to go all in on BizDev — which is a total narrative violation. Nobody says, “Hey, go all in on BizDev.” But that’s what you need to do for this business. What you’ve got to do is create some sort of digital likeness license. You need to go to athletes and celebrities — everybody CAA represents and WME — you’ve got to partner with them and basically say, “Hey, I’d like to sign up to be your rights provider and licensing technology, so that anytime a brand comes and wants to use your name and face and voice, they can do so and you can get paid for it.” They can have the official training data for you. They can have your signature and your rights saying you’re allowed to use my face and my voice to do this.
After you have the rights — and this is beautiful, because we all know in businesses the value is in the defensibility — and the defensibility here is not the technology, it’s the rights ownership. If you can go lock up the rights — there’s a land grab right now — if you can go lock up the rights to the right names and faces and voices, and you own that, you either license it from them and then sublicense it out, or you create the product that they use to manage their rights and licenses. You have a moat, because celebrities only want to do this once. Whoever gets there first wins.
After that, you can either buy technology, partner with technology, do whatever you want to actually deliver this deepfake stuff. In the future, when LeBron James does a McDonald’s commercial, he’s not going to have to fly to Georgia and film that commercial. They’re just going to put in an API request, pay the money, get the license to his face and voice, and then they’re going to be able to script it. He’s going to be able to sign off for approval. That’s it. That’s how the whole process is going to go. He’s going to make more money with less time. That’s where that’s going.
Idea #5 — The Anti-Cameo (Digital Likeness Protection) [00:24:15]
Shaan: Idea number five is the opposite of that — the anti-Cameo.
I was on TikTok yesterday and I saw this video of Mr. Beast giving away iPhones. Now your boy’s not a dummy, so I know he’s not actually doing that. This was a deepfake. But a lot of people didn’t know that. They were clicking on it. In fact, Mr. Beast tweeted this out: “Hey, a lot of people are seeing this and this is messed up. How can I get rid of this? How do I stop this? This is terrible. You’re using my face in ways that I definitely don’t want.”
Similarly, I saw — Taylor Swift is all the rage right now because she’s dating Travis Kelce — somebody used Midjourney or one of the AI image tools to make photos of Taylor Swift smoking weed or whatever. They put her in a bunch of compromising situations that you wouldn’t want to be in. She doesn’t want to be in that. That’s bad for her brand. But now it’s easy for anybody to just deepfake that.
You need an anti-Cameo — the takedown and protection and detection and monitoring service for all these celebrities to say, “Hey, we are scouring the web to make sure your face is not showing up in deepfake porn or in ads that you’re not actually endorsing.” That’s going to become a no-brainer business. It’s a consequence of how good this technology is.
By the way, for all my crypto haters out there, this is a great blockchain use case. There’s a company doing this that basically says, “We’re going to let you create — everyone who is a celebrity, any company — you have a private key. Anytime there’s media out there, you can sign the media with your private key that only you control, to say, ‘Yes, this is real. I actually said this. I actually vouch for this. I actually created this. I actually endorsed this.’”
What’s going to happen is that in the future, media that doesn’t have that digital signature is going to be seen as untrustworthy. What they did with the blockchain was very smart — they let anybody create their keys, their digital signature, so that they can sign off on these and have it encoded that “Yes, I indeed did say this, I do say this, I endorsed this, I allowed for this.” That’s going to be a big deal.
Idea #4 — AI Tutors [00:27:00]
Shaan: Number four: AI tutors.
Education is obviously a big space. How amazing would it be if everybody had their own AI tutor? Everybody had a patient, infinitely intelligent tutor who is going to let them just take a picture of the problem they’re trying to solve. You can ask it to explain it. You can ask for more examples. You can ask it to go slower. You can ask it to go faster. The tutor can quiz you after explaining it to make sure you understand the concept. The tutor can draw diagrams. The tutor can animate diagrams faster than any human tutor could, better than anyone could on a chalkboard.
Ultimately, they can keep track of what you have mastery on and where you need work, and it can customize a curriculum that fits you — versus just you going by the book of whatever the herd is trying to learn right now.
I think an amazing thing that’s going to happen is that we’re going to get these AI tutors that are really going to help teach people. I think this is one of the big use cases of ChatGPT today — sort of like homework shortcuts. But I think part of those shortcuts are going to be not just “do it for you” but “explain it to me if I want to learn.” You’re not going to be able to force people to learn, but for the people who do want to learn, I think an AI tutor is going to be a game-changer.
I know for me, I used to love watching Khan Academy videos anytime I had a question, because Sal is just an amazing teacher. He could teach millions of people because he’s patient, he’s clear, he has a very good little drawing setup. This is what’s going to happen with AI. We’re going to have Sal Khans for all of us. Some that are funny. Some that are good at math. Some that are good at really slow explanations. Some that are very advanced and keep you on your toes. You’re going to have different AI tutors to choose from.
Idea #3 — Call Center Accent Changes [00:29:30]
Shaan: Number three: call center accent changes. This is a fun one.
A lot of customer service help, as we know, is offshore. It’s cheaper — that’s why businesses do it. However, there’s one big cost. People get enraged when you call — you’re calling Dell for help, and all of a sudden you’re getting Arjun in India who says, “Hey, it’s Arjun from Dell,” and he’s got the accent and he doesn’t quite understand what you’re saying, the connection isn’t great.
Well, there are companies right now that are doing AI accent removal. What they’re doing is: the guy Arjun is in India, but when he talks, he sounds like Adam in California. Because on the fly, they’re able to adjust using AI his voice so that he doesn’t have the accent.
Think about it: if you’re a company, would you not pay an extra 5%, 10% to be able to remove accents from all of your customer support people so that you have higher NPS, higher customer service scores, and fewer issues and complaints? Of course you will. Now you multiply that across a very big industry of customer support, and that’s one really specific way that you take this technology unlock of being able to imitate anyone’s voice and you apply it in a business context.
Idea #2 — AI Porn (The Fantasy Factory) [00:31:30]
Shaan: Number two. I had to do it. I had to do it. You know me — I’m the guy who’s been telling you that OnlyFans was going to be a big business for years now. AI porn, a.k.a. the Fantasy Factory.
Porn is one of the biggest markets in the world. Historically, porn has been an early adopter of new technology. Video streaming early on was used for porn. Online payments early on were used for porn. Porn has been an early adopter of many new technologies, and I think there’s a big benefit for AI porn.
On the consumer side — the demand side — you have infinite personalization to your taste. Have you ever gone to a porn site and seen how many categories there are? There’s like a trillion categories. Why? Because there’s a trillion fantasies that people have that they’re looking for. Whatever number of categories they have, there’s probably actually room for 100 times more, because that’s just what they’re able to service — that’s not actually the limitation of what people want or what they’re interested in. Infinite personalization is the first thing. “What exactly is your thing? We can provide that to you.” That’s going to beat out somebody who just says, “Here’s what I have, do you like it?” It’s the difference between going to Blockbuster and seeing 25 movies on the shelf versus Netflix or Amazon which have an infinite shelf. It’s the difference between watching one episode of America’s Funniest Home Videos versus opening up TikTok anytime you want, swiping, getting personalized entertainment, bite-sized, for an infinite scroll. One is a lot more powerful and will satisfy demand much better.
Then the supply side. AI porn, I believe, is more ethical. Just like the way we have plant-based proteins and cruelty-free makeup and vegan leather — no humans are going to be harmed in the production of this video. I think the idea that we could satisfy the demand for porn without subjecting people to the lifestyle of being a porn professional — is that what they call themselves? Porn professional? Porn stars. I like that. I’m a business star, not an entrepreneur.
I think that’s going to happen. For the companies, it drops your COGS down. You don’t have to worry about all the takedown notices and copyright issues and lawsuits from people uploading stolen works, or having to share revenue with all the production companies. It’s just going to drop down to whatever the cost of the GPU is to create the thing.
If you ever want to go down a rabbit hole, go look up the company MindGeek in Canada. They own all the porn sites. It’s a multi-billion-dollar conglomerate just spitting off tons and tons of cash. Somebody’s going to go after that market with AI, and I think they will win, because: way more powerful tool to satisfy demand, way lower COGS, and more ethical supply. That’s a winning formula.
Idea #1 — The Self-Doing To-Do List (AI Agents) [00:35:00]
Shaan: All right, now we’re here. Number one. This is it. This is the winner of my list. This is the number one biggest opportunity.
I hate to be that guy that’s like, “This is the biggest opportunity of our lifetimes.” Right? Like the timeshare guys who are now YouTubers just selling the biggest next big thing. But in this case, I think it’s actually warranted. AI is a big, big freaking deal. You got to be nuts to be denying that at this point. And within AI, this is what I genuinely believe is the biggest opportunity.
Let me just rewind for a second. If you think about the previous waves, I count four waves in my lifetime of giant inflections or opportunities — the big gold rush moments, the huge tidal waves that you could go surf as an entrepreneur.
The first one was early internet. You had what they used to call the “information superhighway” — if you remember that dorky phrase — but it was right. Information was the thing. The ability to find anything. There are really two big winners: Google and Amazon were the two biggest winners of the information “find anything” paradigm. Google let you go find any information. Amazon let you go find any stuff. Those two created a trillion dollars, more than a trillion, almost $2 trillion of value just on those two companies alone, out of the information wave.
The second wave was communication. Started with email, but then quickly became all social networking, social media. Facebook, LinkedIn, Twitter, Snapchat, Pinterest, WhatsApp, YouTube — all of that. The ability for people to communicate with each other. This even extended to like Uber — riders communicating with drivers, basically the ability to say, “I need a ride” and somebody else says, “I have a ride for you.” Connecting the dots. Human coordination and communication was the big second wave. Again, a trillion dollars plus of market value created out of that communication wave.
Then what came next? The value wave. We’re talking early 2010s up to basically the last decade, 2010 to 2020. This is crypto. Bitcoin, Ethereum — just those two alone are a trillion dollars of value that got created. The best investment you could make since the year I graduated, 2010, would have been into Bitcoin and Ethereum. Those were the best-performing assets that were created. So we had digital information, then we had digital communication, then we had digital value — the ability to have digital money and digital value.
Now what’s number four? Coming into the 2020s, we’re going from 2020 to 2030. What’s going to be the next one? I believe it’s digital intelligence. That’s AI.
Who’s going to be the Bitcoin? Who’s going to be the Google? Who’s going to be the Facebook of the intelligence wave? Well, so far we’ve seen Nvidia make chips — they’ve done extremely well. We’ve seen OpenAI become almost a $100 billion company by training models that you can use for AI applications. I think those are great, but I think there’s going to be more.
The one that I think is missing — I had two I was thinking about. The first one was self-driving cars, because driving is a huge part of human life. When that goes self-driving, you’re going to have safer, more efficient, more comfortable, more entertainment — you’re going to have a better ride experience. That’s going to obviously change the way that cars work. Cars are parked 90% of the time, but when they’re self-driving, you’re going to park your car and say, “Go make me money, car,” and the car is going to go drive around and be an Uber for people, picking people up. It’s a game-changer in terms of how this works. We’re going to need less cars, which means less parking, which means different roads and city structures. There’s a whole bunch of things that can change.
But as I was thinking about that, I thought about a bigger opportunity. The bigger opportunity is actually related. I was like, “Oh man, self-driving cars are going to be great because you just get in the car and just say your destination, and then the car is going to figure out all the things to do to get you to your goal.” I just say, “Hey, take me to Starbucks.” It’s going to say, “Great” — look up the nearest Starbucks, find the address, input the address to navigation, turn the car on, shift gears from park to drive, accelerate, stop at the stop sign, signal, turn, shift lanes, exit the highway — whatever. All of the middle steps. It’s going to create a list and it’s going to do them.
What I realized was that’s not going to stop at just cars. The biggest opportunity is the self-doing to-do list. Just like we have the self-driving car, what’s a self-doing to-do list?
All of our productivity today comes from basically the following system: person thinks about what they want — goal. Then they create a list of actions that they think will move them in the direction of that goal. And then they do those actions. The extent to which you hit your goals is based on your ability to know what you want, create the list of things you need to do to get there, and then actually do the list.
I think what AI is going to do is actually change how that works. It’s actually going to let you just say what you want. AI will then generate the list, and then it’ll just do it.
We’ve seen this, by the way. What this is called right now — they’re calling these “agents.” Agents might end up being the “information superhighway” — a word that gets phased out over time. But here’s the simple model. We’ve all used ChatGPT where you type something in and it gives you an answer. The better question you ask, the better answer it’ll give you. You have to know what to prompt it to do. If you don’t do anything, it’s just going to sit there and do absolutely nothing for you.
There’s a new model called AI agents. What AI agents are based on is that you don’t need to ask specific questions or give specific instructions for a task. All you need to do is tell the AI what you want as your goal.
Let’s just take an example. “Hey, I run an e-commerce business and I want to reduce my inventory waste. I’m going to make my inventory more efficient.” So the AI could then generate a list of things to do: analyze the inventory to find the highest fast movers and slow movers. Then it will take the slow movers and put them on sale. It will take the fast movers and analyze where there’s more demand than you have supply in stock. It’ll create a purchase order, send the purchase order to the factory, and get the next order delivered for you. You can see, in theory, how you’d be able to just say a goal and have the AI create a list and then do it.
Another example — let’s say I wanted to lose 15 pounds. We all know to lose 15 pounds, you need to burn more calories than you consume. In theory, AI is going to be able to help you do that. You’re going to say, “I want to lose weight,” and it’s going to say, “Great. We are going to take your current weight, create a calorie meal plan of how many calories you’re supposed to intake per day and how many you’re supposed to burn per day, which is going to create a workout plan for you, a meal plan for you. It’s going to take the meal plan, break that down into recipes, break the recipes down into ingredients. It’s going to go on Instacart, order the ingredients to your house for you.” There you go. It’s going to take you as far as it can.
It’s not going to be able to do every single thing in the real world until you have a robot sitting in your house that’s going to take those groceries and prepare the meal for you — Jetsons style. We’re talking about the future, but this is the future.
I mean dude, I used to have this — I’m holding up my cell phone right now — that I can use to run my business, to entertain myself endlessly for hours, to play video games, to navigate all around the world. I can pay for things on this. I don’t need my wallet. It’s insane. This thing is insane to 12-year-old me. And 12-year-old me is not that long ago — that’s 20 years ago or whatever. 12-year-old me would be mind-blown because 12-year-old me had just gotten their first computer in their house. We used to have a computer room. It was a room in our house called “the computer room,” and whoever wanted to use a computer had to go to the computer room.
When you were in the computer room, you got the internet through a CD from AOL. And when we had that, you used to have to pick and choose: do you want to be able to receive phone calls as a house, or do you want to be on the internet? Because if somebody picked up the phone when you were on the internet, you would disconnect from the internet and they would hear crazy internet sounds. It was insane — the way we were when I started on the internet to what now, holding my iPhone 15 Pro Max with wireless internet while I’m driving or on an airplane. Mind-blowing.
Closing — Start the Bonfire [00:42:30]
Shaan: All I’m asking you to do is just sort of think 20 years in the future. The idea that we’re going to have our Jetsons robot in our house, and we’re going to be able to tell our to-do list just what our wishes and dreams are, and then it’s going to create the list and do them — that’s the big idea. That is the big idea. That is the idea that is mind-blowing.
That’s also the idea that scares people, by the way, because the thought experiment is: someone says, “Hey, I want to maximize” — you know, the paperclip example — “I want to maximize the production of paperclips, or sales for my paperclips.” And the AI got your goal and it doesn’t care what comes in the way. It’s going to start shredding cars to create scrap metal to produce more paperclips. It will do anything to hit that goal. That’s the scary part about AI.
Now I’m not the guy to do the AI safety conversation and the AI ethics conversation — that’s not me. I’m an idea guy. I’m thinking about how technology can do really cool things that will improve people’s lives. And then of course, as we do them, we’re going to need to put in guardrails and guidelines and be able to not crush all of civilization in the process. But that’s kind of a downer. I’m not really looking to get into that conversation.
What I’m excited about is a future where work gets done for you. That’s what’s going to come from the intelligence wave. That’s different than the value wave, the communication wave, or the information wave. The intelligence wave is going to do intelligent things. It’s going to use its brain for you. It’s got this right side of the AI brain that can do creative — it can draw, it can write, it can sing, it can make songs, it can rap, it can do anything. And then on the left side, it’s got this informational, analytical brain — you dump in a PDF and it’ll summarize it for you in a second. It’ll generate a P&L for your business. It’ll give you strategic advice on your taxes. It can do all these things that are highly left-brain.
Now we have this AI intelligence brain, and the ideas that I came up with were just my first pass at what’s going to come from this. I want to hear what you’re going to do. I mean, I’m investing in this space. If you’re doing something cool, reach out to me — shaan@shaunpuri.com. I want to hear what your ideas are, and I want to hear what you think about these. Go in the YouTube comments and let me know.
Those are my ideas from 10 to 1. I hope you liked it. I hope you enjoyed it. I hope I started a little bonfire in your brain. All right, I’m out of here.