← Digital Disruption with Geoff Nielson
Digital Disruption with Geoff Nielson
Boom or Bust? Top AI Investor Reveals the Future of AI Startups
01:01:26
Loading summary
Transcript84 lines
- [00:01]
A
Hey, everyone. I'm super excited to be sitting down with AI venture capitalist Jeremiah Aoyang. He's the general partner for AI investments at Blitzscale Ventures, set up by the founder of LinkedIn to bet on tomorrow's AI leaders. What's cool about Jeremiah is that he doesn't just follow the trends. He's the one deciding which AI startups get the chance to change the world and which ones don't. I want to know what makes him tick. What does he look for when he's trying to find the billion dollar technology, which cutting edge technology is getting his attention and where he thinks all of this is taking us next? Let's find out. Hey Jeremiah, super happy to have you here on the program today. You're the general partner of AI investments at Blitzscaling Ventures. First and foremost, I guess I just wanted to ask you, you know, you're an AI startup investor. What does that mean? And what type of businesses and startups are you looking at in 2025?
- [00:58]
B
Well, Jeff, first of all, it is a delight to be here. It's a wonderful day here in Silicon Valley. I'm just immersed in the Silicon Valley ecosystem and I got to tell you, from the macro view, it's all about AI. So the main freeway 101 is around 80%, 75% of the billboards are about AI. And there's typically five to eight AI startup events per night. There's typically two to three to four hackathons each weekend in San Francisco and in Silicon valley, each with 300 developers building. I've been in Silicon Valley Since.com era and I've never seen the level of energy here. So my job is easy, but hard at the same time. As a vc, it's easy because the startups are here in Silicon Valley. But the hard part is there's over 30 engineers, 8,000 AI startups and projects according to the database. There's an AI for that dot com, of course. And it is just an amazing time to be here in the early part of this journey. And I've been in the generative AI this part of the market full time here as at the VC firm for two years, made eight investments and we're looking for early stage startups like Pre Seed Seed and sometimes series A startups that are starting to show growth and traction. And there's a couple of things that we look for but at a high level. That's what I'm seeing here, right, right in the heart of Silicon Valley.
- [02:26]
A
It's so amazing. And the volume is just as you said, it's hard to even fathom that much excitement, that many people, all with the same energy toward AI. One of the questions I want to ask you before we go deeper into the nature of the startups themselves is you talked about.com and having obviously there's a long history of boom and bust and hype in the Valley. From your perspective, we hear so much about AI. Where do you see us as being in the hype cycle? Are we kind of over a hump, so to speak? Is it just going up from here? What does that look like from your perspective?
- [03:05]
B
It's challenging to tell, but I still think we're in market formation mode, which is towards the early part of the cycle. So I've been through five tech Cycles.com, web2, the sharing economy, Web3 and now AI. My background, by the way, I was an industry analyst at Force to Research. We spun out, we started our own boutique industry analyst firm which my job was to forecast what's going to happen in tech and business. And I've been a professional speaker and then an angel investor which led me to a full time role at a venture capital firm. So tracking trends is really part of my DNA and what I do. So we're definitely in the early point of the market. You know, again, 38,000 startups, it's like the Cambrian explosion of all these startups. There will be a point in the market where we will have consolidation, there will be acquisitions, startups die and we're seeing some of that happening now. We're seeing some acquisitions happening now and so that's just starting to show. But we're, we're still in phase one as market formation. This feels like a 10 to 15 year cycle, but it's really hard to forecast. Even our advisor of Blitzscaling, Reid Hoffman, he says he can only see two years out and he's the first investor in OpenAI and he's on the board of directors of Microsoft and he's guiding us. Our firm is based upon the book called Blitzscaling that he wrote with Chris Yeh, my friend for 20 years. So if Reid Hoffman can only see two years out, you know, I'm definitely going to be able to see a lot less than that.
- [04:35]
A
So, so let's talk about that scale for a minute. So if we think about two years out or know even six months out, what are the big trends or the big patterns that you're starting to see emerge that you think are really going to take off over the next handful of months.
- [04:51]
B
So let's start with today. Right now, the topic du jour is AI agents. And AI agents complete actual tasks. Large language models are great at completing sentences. It's text. But AI agents actually will go into your inbox, manage your emails, can go and help edit, you know, a video, can create, you know, workflows, can actually talk to customers, do customer care, do marketing for you. It actually completes the tasks similar to entry level workers. Most of any entry level task that is on a digital device that will be automated within the next two years by AI agents. That's pretty much a given in Silicon Valley. We all expect that to happen. In fact, many of the startups have 22, two employees and they're generating millions of dollars revenue per employee. It's called a lean AI native startup and they're using AI agents. So that's today at the front of the market, AI agents are helping people to get things done. Looking forward into the future, there's a number of trends to look at. One is that we're seeing AI agents in deep verticals. So we're going to see agents in healthcare, agents in finance, agents in real estate, agents in even government, AI agents in military. Another trend we're seeing is called world models. World like you know, the planet. And just as we've seen large language models be able to predict what human sentences, how they will be completed, that's essentially what they do. A world model will predict what you're going to do physically. Like humans are actually very predictable. Like we go to work, you go to the bathroom, you get coffee, or maybe reverse that order, you're going to go to lunch at 12 and you get in your car at 5, you go home, you're going to go to this store versus that store, we know what radio station you're going to listen to. So a world model predicts what humans, animals, weather, cars, those patterns on what people do. So that's a new emerging category called world models. Now after that is called humanoid robotics or the physical instantiation of AI into the physical world. We shift into self driving cars and robots will be among us and humanoid robots, so that's coming as well. So those are just a couple of trends that we're tracking and then which I will be investing in. So it's my job to think about what's going to happen and make sure I understand the market, have the right connections and know the founders, make friends with them, show them how we can help them as in, and then invest in those startups. So that's kind of how I see the future rolling out.
- [07:29]
A
It's, it's. Both of those examples are super, super interesting to me and you know, clearly big gnarly problems to solve. Are you starting to see with some of the startups that you're looking at and looking to invest in, are some of those starting to be in that space or is it still predominantly in kind of the AI agent and generative AI space that you're primarily focused?
- [07:54]
B
We've done a number of investments and around half of them are AI agents, the other half are in specific verticals like I already kind of mentioned like how we think of the market and so that's here now and I'm exploring the world model space and then the robotic space is next to understand like how that's going to sort out what I do to stay on top of the market is. So I, I attend on average three AI events in Silicon Valley SF per week. So that's like 150 events per, per year. And that's just, I love it and it's, I've always been doing that here in Silicon Valley. So I'm just hyper connected. And when I go to an event I know about a third of the people in the room just because I show up so much and then another third of the room I can get introductions to and a third, I don't know. You know, people fly in. Right. So usually I have enough network where I can get all the signals and meet people or get to people I want to talk to and, and listen and learn and, and do things. I host Llama Lounge, one of the longest running Silicon Valley events and we host it several times per year here in Silicon Valley. And typically it's 300 AI founders, CEOs that show up and around 75 VCs. My network, my friends, my peers, sometimes competitors by the way, and then around 50 Fortune 500 executives and AI leaders. So I'm just hyper connected to all of these things. And for sure I'm seeing this rise of robotics emerging with the AI space that gets all, starting to converge at a high level and I'd love to hear your thoughts is AI is permeating every industry and every technology trend. It's like it's everywhere. It's, it's just kind of like air where we're just breathing it. It's connected to every single technology movement.
- [09:40]
A
So Llama Lounge, you mentioned that it's hundreds of AI founders, it's VCs, what does it actually look like on the floor? And in practice, is it tech demos? Is it pitches, is it conversations? What actually transpires at an event like that?
- [09:53]
B
Well, I'd love to invite you. And next time you're in town, please be my guest. So it's hosted at one of the top tech companies, one of the hyperscalers. So it's been hosted at Amazon Web Services, SAP, Capgemini, Stanford Campus, Zoom Campus. The mayor of San Jose has shown up. The founder of Zoom, a billionaire, Eric Yoana, showed up. Press and media. It's been covered. The Atlantic, the New Yorker, TechCrunch. So it has really grown to a massive thing. And you can see the group photos. We take one giant photo. So what everybody usually says is that the energy level is so high, people are so excited to be there. We decline. Hundreds of people. There's usually a line down the street to get in. There's videos showing that happening. And what happens is, out of the 300 to 400 people that attend, many of the founders are applying to demo. And there's 10 demo slots that are offered to these startups that I choose. And by the way, three of the 10 slots are always reserved for female founders. And we want to make sure. And often we're at parity, and it's really important we have that diversity and we try to ensure that we see all the different things. So there's demos on the floor, and then we all come together for the kickoff, and I do something very unique. So there's usually 100 to 200 AI events in Silicon Valley per month.
- [11:06]
A
That's.
- [11:07]
B
It's insane. Right. So how do you stand out? Well, we got a great brand, we've got a great logo. The who's who shows up. But what I do to make it unique is I have a giant conch shell. Yes. And I'm a former jazz trombone player, and so I play it and everybody knows it and they scream it. So it has its own audio signature kicking off that event. And the energy just goes up. Boom, boom, boom. All these notches. We have a keynote speaker and then we have the 10 startups who come and pitch for only 30 seconds. It's kind of like that Eminem song. Like, you have one shot and they got one shot to pitch in front of 75 VCs and 300 CEOs. And this is their shot. And so this is like a big moment for them. And many of them get funding, customers, partners, employees from that event.
- [11:53]
A
Wow. No, it's, it's, it's amazing. It's like, it's, you know, sounds like kind of Like Shark Tank on, on speed or something. So it's really cool to be able to see that level of excitement and that level of investment. So I want to come back, Jeremiah, to one of the questions I asked earlier that we started to get into, which is, you know, for you and for some of these VC investors, what are you looking for these days in, you know, in an AI or in a tech startup? You know, obviously blitzscaling is kind of the name of the game, but what are some of the indicators that, you know, there's, there's the right DNA here, whether it's the right people or the right idea.
- [12:31]
B
Great. So it depends on the stage of the startup. So an earlier startup, we're going to look more carefully at the founder, his or her background, what is their pedigree? Do they, can they learn quickly? Do they have a track record of doing new things? What's their attitude and why are they really doing this? Like really getting down to core motivation. So that's really important in the pre seed and seed towards series A or B, the company scaling, sometimes the CEO is even replaced by a professional executive. Right. So at that phase we're looking at the scaling attributes all throughout our thesis because of the Blitzscaling book, which used to be a Stanford class, where there's two major business strategies. And first, I want to just share that a technical moat in AI lasts only a few months before it used to last years, no longer. In one tweet from Sam Altman, your company could get wiped out.
- [13:30]
A
Wow.
- [13:30]
B
Literally, it could be overnight, right? They'll announce a new feature, the one you've been working on for two years. You're ruined. So a technical moat is very thin, inches, feet. It's not a very wide moat and certainly not deep. Very few people have a real technical moat. If you do, it's not going to last long. So what we look for are a few other business strategies. For example, do you have deep industry experience in a specific vertical? Do you have all the contacts in the construction space? And you know that space because OpenAI is going horizontal, Anthropic is going horizontal. Google, Microsoft, Amazon, they all go wide. So if you have deep expertise in one vertical. Secondly, do you have proprietary data that nobody else has and that should be a legal contract with clients, or you've acquired IP and you've trained your models or agents on that specific thing that nobody else has. The key classic business strategies to blitzscaling, there's two major things, and this is based off the interviews that they did at Stanford with the leaders in the market, it's viral effects and network effects. So a viral effect is when the product spreads on its own with and we can look at this data, low or no marketing, low or no sales, low or no advertising. So if you strip away that data, by the way, I'm not saying don't do those things, I'm just saying let's pull the data away. Does the product actually spread organically? Whether it's word of mouth, more importantly, product led growth, where the product is designed to spread. DocuSign, Box, LinkedIn, Airbnb, Uber, they all do that. And three, the third thing we look for is, is there a distribution strategy? Like for example, does Salesforce promote your SaaS startup in their network? Do you have a lot of GitHub stars and therefore everybody sees it and it spreads. So that's, that's the key strategy is viral effects. Does the product spread on its own? The second one we look for and we have a scorecard, we score companies on this, they have to have 80 points or higher in total. If so, we bring it to Reid Hoffman for a review. The second one is network effects. Network effects are when the value increases as every new user and customer joins. So Airbnb, Uber, Facebook, social networks, marketplaces, exchanges, app exchanges, B2B cloud infrastructure, they all have network effects. The more people that participate, the value increases for everybody else. So viral effects are key because the product spreads to the whole market at low cost. But then you gotta keep the customers. So network effects means the value increases, therefore increasing the loyalty and reduces your switching. So that's what we look for. Awesome Founder Deep Vertical Knowledge Proprietary data Viral effects Network Effects reviewed by Reid Hoffman.
- [16:28]
A
I love that model, Jeremiah. And it's so clear and so rational to me. The one piece I wanted to ask you about, so viral effects, product led growth, all of that like completely computes in my mind and makes sense. Network effects to me seems like it could be a double edged sword, right? If it's saying like, oh, in the future if everybody joins this, it's going to be amazing, but right now it's not. How do you mitigate that risk? Is it only joining at a certain phase in the growth or how do you engage with that?
- [17:02]
B
Yes, great point. So at Pre seed it's difficult. Oh, sorry, it's more difficult to determine do they have viral effects and network effects in pre seed, the early stages of the company. But right around seed they're starting to get traction. The product's built, they're starting to get product market fit. They kind of know who their ideal customer profile is. So we can see their business strategies by then. So it's pretty obvious. So some of the companies, they already have a data network effect or the AIs learning that is a network effect, or they have multiple customer cohorts that are collaborating in a platform that's a network effect. We can see that in the product already, often in seed and often in the pre seed, the roadmap. I don't tell them too much what we're looking for, but they should be smart enough to just like perplexity it like network effects. I can see it in the roadmap if it's coming. So those are just some of the kind of things that we look for, even though it may not be fully built out.
- [17:55]
A
Got it, got it. No, that makes complete sense to me. One of the things I wanted to ask you about, and you already started answering, is just how you see the shape of the AI marketplace and the AI industry kind of unfolding. And you mentioned that for a lot of the big players like OpenAI or some of their peers, they're kind of horizontal and fairly shallow. And you're looking for more deep vertical plays in specific industries. Is that, I mean, first of all, is that a fair summation of your prediction of the shape and is there any more nuance either in industry or geo or any sort of demographics that help you predict and understand what this could all look like a few years from now?
- [18:42]
B
Yes, so we have done some horizontal bets, but they're not in the spaces the giant foundation models or the giant tech scalers like Google, Amazon, Apple are playing in yet, which is fine because they are likely to acquire these startups. So we definitely done some emerging horizontal plays and then also these vertical plays. But it's too late for me to get into open AI or anthropic. Right. Those companies are worth billions and trillions now. So in general, I see that now this is important to note that many startups that are around the world, they are often very successful in their specific region. And you asked me about geography. I'm in the Middle east very often. Um, I've been in Asia quite a bit. And they are great at tackling their market, which could be a culture or a language or a region or a country or a specific area. And one of the requirements of many, maybe most VCs in Silicon Valley is that the startup needs to go global. And so that's often a big challenge for regional based startups. But the founders often fly to Silicon Valley to meet the, the ecosystem and to raise money and to show their plans on how they plan to go global. So that's an important thing. Now with that said, we've invested two times in India, one time in Canada in the fund that I'm managing and we'll continue to invest globally. But a mass majority of the startups, oh, in Brazil. But the mass majority of the startups tend to be headquartered in Silicon Valley. There's just the acceleration, the movement, it's just happening so fast. At my own event, Llama Lounge, I poll the audience to see where you from. The mass majority are from us and then live in Silicon Valley. But the second largest group, which is a 20, 25% are Europeans. They're coming here because they can innovate faster. European regulations have been a stranglehold for many of them.
- [20:31]
A
That's interesting. If you'd asked me to guess where the second largest group was from, I don't know that I would have guessed that, but that's really interesting to me. Yeah, one of the kind of adages about AI and AI tech, big tech players you hear these days is that we're in an AI arms race and it's going to be winner take all and everybody's dumping money into it and that the winners are going to be kind. The people seem to think it's going to be one of the incumbent big players in the space. Do, do you buy that? Like, do you see the, the big tech players as having a stranglehold on it? Do you see new players emerging and what are the implications, you know, as it pertains to some of these startups, do they just need to get acquired or is there room to kind of grow on their own?
- [21:17]
B
Great question. So I published a diagram called the AI Tech stack. I also did one called the AI Agent ecosystem. It's a framework, you can just do a perplexity on those and you'll find the graphics on how I've literally drawn out the market in boxes to show the segmentations of the different products and the different companies and how they'd segment again. My background is I'm a professional industry analyst and forecaster turned vc. So that's a natural skill to me. And it depends on which part of the market where we, we will see a winner that takes most. Now we're often asked, I mean this is a complex question question. So whoever gets the quote AGI first is going to have a dominant position, but it will probably be multiple companies that achieve that. And in the a, in the era of AI agents where you will use your agent app to get most things done. You don't have to visit websites in the future, you don't have to visit apps. The agent will do those things for you and they will also go to podcast and listen to it and then and report back to you the way that you want to consume that information. Like, you don't have to leave that experience. We're going to see a big shift in how we consume media now. With that said, I don't think we'll have one app. That's your AI app. My thesis is, and it remains to be seen, is you will have the equivalent to the current amount of email accounts that you have now. So, Jeff, how many email accounts are you managing?
- [22:42]
A
Oh, it's a good question. Too many. Let's. I don't know, more than two or three, let's put it that way.
- [22:50]
B
Okay, cool. So let's say three. So you'll still have three very powerful AI platforms and therefore companies in your life. And there's a lot of reasons why, is because Apple vs Microsoft vs OpenAI vs Claude vs Amazon Apple, they're going to fight tooth and nail to make sure that they still have a piece of your attention or they have a piece of your data and they're going to give you great options and competition is wonderful. And so I think that's going to continue to happen. So I think I have three to five email accounts that I'm actively managing, managing. So that's how many dominant, what we call the meta agents, by the way, that will be in our lives. So I don't think it's one winner that takes all.
- [23:33]
A
So I like that model. It resonates with me and it's in line with what I've seen, you know, both in kind of the consumer and the enterprise space. And for a lot of what we've talked about, it's been the consumer space and you know, we talked about vertical in terms of the enterprise space with your bets and the AI, the AI startups you're looking at, is it predominantly consumer, is it predominantly enterprise or is it, you know, kind of a blend?
- [24:00]
B
It's a blend. So I have an enterprise background. I was at Hitachi and data centers and obviously industry analyst is very enterprise focused. So we've bet on a number of enterprise players. In fact, 95% of EC funding, I believe, goes to B2B, not B2C. B2C is. The chances of success are minuscule. It's very hard market. So most investment goes to B2B.
- [24:25]
A
Right.
- [24:25]
B
So we've invested in one company which is very popular and growing fast, is called crewai. Open Source AI Agents for the Enterprise. And I want to just share this concept of the lean AI startup. Have you heard this or had any guests talk about this? It's a concept that's been growing in the last few months in Silicon Valley. So a lean AI startup has the AI first mentality where you, if you have a business problem or a personal problem, you first see if there's an off the shelf AI that exists or an agent. And if you can, you grab it, you download it or you use it. And if it doesn't exist, then you build it. And there's many no code tools that can help you do that. And if you can't do one or two, then you hire somebody. So right now Silicon Valley is obsessed with this lean AI native leaderboard. It's actual, it's a notion page and it lists out around 40 startups. And the requirements to be on this leaderboard is the company has to be less than five years old, it has less than 50, fewer than 50 employees and it's generating $5 million or more. Now this is the shocking stat. The average SaaS company generates around US$200,000. ARR. That's revenue per employee per year. 200k. Just hold that number in your mind. 200k average employee revenue. You know the big tech companies, the big SaaS companies are earning that. The lean AI native startups coming out of SF like 66%, 70% are from SF. They're generating 2 to 3 million dollars per.
- [26:03]
A
Wow.
- [26:04]
B
We've never ever seen that. So we invested in crewai, which is doing hockey stick and adoption. There's, they're, they're following this, these practices, these methodologies. There's four shocking stats with Crew. Crew is in 60% of the Fortune 500 enterprise agents. Number two, the company's only, it was birthed around two years ago. Less than two years ago there's 30 employees but 300 AI agents that are doing the work. So that's the future of companies. 10 to 1 AI agents per employee. Like today in this year, you know, that number will increase agents per employee over year and revenue should go up. So that's what a modern company is looking like out of Silicon Valley. Two to $3 million revenue per employee, we've never seen that before. That's 10x revenue compared to traditional one.
- [26:58]
A
It's especially cool to hear that CREW is actually doing that and it's not Just, you know, they're promising that they'll do that one day. And the reason, Jeremiah, why, why that's so exciting to me is, you know, I've had this hypothesis for a while now around AI agents is, you know, I think a lot of people out there probably heard about this for the first time from Marc Benioff, you know, getting up at, you know, Dreamforce and basically Agent Force. Yeah, yeah, yeah, Agent Force just you know, agents, agents, Agents and you know, a version of like the ballmer develop. Developers, Developers. And it's easy enough for me to imagine a world where the big B2B enter Enterprise players like Salesforce just have a total stranglehold on the agents that organizations use. And it sounds like you're saying that doesn't have to be the case and it doesn't look like it's going to be the case and that, you know, smaller organizations have, have a shot at this. Is that fair? Am I representing that properly?
- [27:55]
B
Absolutely. So the big enterprise companies like Salesforce, Adobe and Amazon, they're big, but they're also very slow. So the startups are maneuvering much faster and getting faster adoption. Plus open source offers something that these lock in SaaS companies can't offer and many CIOs are really tired of being locked in. So open source is a faster way and they can control the data, they can download it, use it on prem. So there's numerous reasons why the startups are playing here. Now Salesforce doesn't have a lock in at the same A week before Agent Force, HubSpot launched Agent AI, Dharmesh pushed that product out, which is a marketplace of AI agents. And now every other big company in SaaS and B2B is launching agents of some form. So it is going to be a standard offering by the end of the year. In fact, most announcements happen in fall, so we should expect to see that happen across the entire industry.
- [28:51]
A
Yeah. Wow. So yeah, it sounds like in one form or another the agents are here and they're starting to have an impact on organizations of all sizes. Jeremiah, I wanted to ask you about that impact specifically and who you see as kind of being most at risk of disruption from this. And there's all sorts of different answers from more junior employees to more mid level staff to, you know, the types of organizations that are going to be disrupted and how it changes the industry. What are some of the macro patterns that you're starting to see emerge?
- [29:27]
B
Thank you. So we should first start with some humility. So when I speak at conferences, I ask the audience, raise your hand if you think people, people around you in your organization will be replaced by AI. And of course everybody raises their hand. Then I say raise your hand if you think your job will be replaced by AI. And only like 5% of the hands go up. So what's going on here, friends? We gotta be honest. Everybody's at risk. Unless of course perhaps you're in giving, you know, bedside manner like a nurse, right. Or you're in a market where there's more demand than there will ever be, which it's just hard to see like that's going to be impacting the mass. Majority of people should be at risk and even VCs are at risk from AI. Everybody's at risk. So I think we need to understand that if you work behind a digital device, mobile phone or laptop or a tablet, your actions are being recorded and therefore they can be reproduced. If your actions physically are being recorded on a camera, that will feed the world models. And eventually we will see robotics in the prime industries. Like we're already seeing that in dark warehouses. It's happening. The most common job code in the United States is a professional driver or trucker. Whereas self driving cars are common, common, common, common in San Francisco. It's a normal thing to see. So there's significant impacts coming to society. So the roles, I prefer to think about the tasks. So the tasks that are replicated and they're of high value, those will be automated very fast. So those are most likely to happen. Now people, there's a big question around, will developers be automated? There's still a high demand for developers and we've seen, seen some of the salaries being offered by the big tech giants. 10 million dollar packages, up to 200 million dollar packages over four years. Obviously the computer scientists, the actual inventors, there's still a massive shortage for the top talent. So it just depends where you look in the market. We probably are seeing already an eradication of customer care, Tier one customer care. I've seen so many startups replacing SDR sales, BDR sales that's already being replicated. Content, content marketing, the communications agency, media industries are generally shrinking. So I'm definitely seeing those impacts happening right now.
- [31:56]
A
So what's the, I guess for us as a society, whether we think about education or workforce, what are some of the implications here? What do we need to be thinking about both individually and collectively as we go through this period? And I don't know if inevitable is too strong a word for you Jeremiah, but it feels like, like there's Almost an inevitability to this disruption to the labor force itself.
- [32:21]
B
I think one thing that to think about and I have a specific question I ask AI leaders and this question gets to the heart of it. And you might want to think about using this question with your experts as you ring across the show. The question is this, how are you raising your children? So I ask that to AI CEOs all the time. I ask that to AIVCs all the time. And they tell me what my child to learn how to be a leader or learn to have empathy so you can influence others or understand what people need or community. The fact that there's 200 AI events in person in Silicon Valley right now, even in the summer, is an indicator that people want to be together, even those that are building AI. So there's actually, there's opportunities if you learn how to look like what those future walls will be and they won't be replaced. So that's the question I ask people, how are you raising your children? So those are skills, right? Or it's not a specific hard skill that I think like 10 years ago we said every child needs to learn coding. I think that's helpful. But it's also going to be useful to know Spanish. Right? You know, I just not sure that there's one thing, one specific skill that you have to do. So learning how to overcome obstacles and be resilient is key. So I have children every year they do a Spartan obstacle race. It's like a military obstacle race out in the dirt and run. I've done over a dozen of them. And so my kids do it and I cheer them on and you know, they train for it. And there's a different obstacle, there's a wall, literally a wall of different sizes. You got to go over it, under it, through it, like whatever, get over it. You got to do it. And so they have to learn or there's monkey bars. And that's a way to teach resilience. How do you overcome obstacles in your life and be ready to adjust this next generation? Generation Alpha, Gen Z is already in the workplace. Generation Alpha is the next they. So I was, I'm digital native. The Internet came when I was in high school, college. I'm Gen X. But Generation Alpha is going to grow up with AI. So they're AI native. So they're going to be used to getting information and recorded experiences from AI and in some cases, like permutations of wisdom that's never happened at that level before. And the AI will start to know their environment and give personalized guidance to them. That's just not something we've ever seen before. And of course, with every technology going back to the advent of fire, right, or steam engines, or the industrial revolution with coal plants creating lots of products, but then we have emissions issues, right? Every technology set has amazing upsides, but also almost always equal downsides, although we're typically in a better place in most cases. So with AI, we will also have great upsides and there's going to be some drawbacks, but we don't really know what all the drawbacks are to society yet. But being connected to AI all the time will help us in many ways, but will also cause us to be dependent on AI. And if it guides us astray, of course that's going to be an issue. And that's a big political topic. Even this around the world right now.
- [35:35]
A
So the AI Native piece is one that's really interesting to me. And the piece that I've been reflecting on lately is just, I guess, the power and maybe in some ways the tyranny of expectations, right? When you've got that AI Native piece, you can get the answer so quickly. And any services that are plugged in, there's such minimal friction to getting you what you need. What does that mean for you? Described earlier enterprise organizations as being kind of slow and lumbering. And I think the two of us, and many people listening to this know that there's an awful lot of slow lumbering enterprises out there that are kind of burdened by legacy systems and processes that are customer or patient or student or citizen facing that are just really cumbersome and slow. Is this wave of technology going to make them extinct? Is it going to force them to adapt? You know what, what's kind of your advice for organizations like this to get to make sure that they're not, you know, the losers of the future and are instead the winners.
- [36:53]
B
It's a big question. So some of those organizations, if they're government, there's no competition, right? There's no, I only have one government where I live, you know, I can't choose a different one. There's no capitalism around that. It's just there's one government, so there's one dmv, like I don't have a choice. So they can go at the pace that they want. And it's just, you can't stop that when it comes to like healthcare, that sometimes we have choices, not always. Sometimes we're locked in. And you know, banks tend to be slow as well. And in some cases it's good that they are slow because they're keeping important information or making life changing choices. They got to be careful and cautious and reduce risk and I'm okay with that. But there's other industries where we will see AI competitors emerge and they will be a direct threat if they don't move quickly. In fact, even in the enterprise technology space, the SaaS companies, traditional SaaS companies. Box. Let's talk about box. For example, that's a rather large traditional enterprise company that lives near me, I live near the office. And the CEO Aaron Levy is focused on AI first even though his company is like over 20 years. Right. So you can see they're trying to change that, the culture as fast as possible. And he's talking about that constantly on LinkedIn. So the tech companies and traditional, they're changing as fast as possible. And the other industries like consumer and auto, they're going to be the second movers. Hospitality and then the, the laggards will be finance and healthcare and then government. And that's a standard rollout in adoption that I've been tracking in enterprise and I don't expect that to change anytime soon.
- [38:28]
A
So with that in mind, I'm curious on your perspective because you probably have a more inside view than most. What you've described in some cases can be really done with a lot of integrity and in some cases it can be an example of what you've probably heard is described as AI washing and just trying to slap an AI label on everything. To what degree are you seeing either? And you know, do you have any tools and tricks for being able to distinguish one from the other?
- [39:01]
B
Yes. So I review 100 applicants per week to my event Llama Lounge, that's average. And I'm reviewing dozens of companies that startups that apply to demo and then literally as I'm driving or an Uber or in a self driving car, I see the billboards here in Silicon Valley of the large tech companies that are AI washing. And it's very obvious when you visit their website, they, they've added GPT so you can have a natural language process with their, with their, their data. It's really not impressive and innovative at all. So you can tell pretty fast like who has something that is just a AI wrapper versus somebody who's integrating it like AI native company. So I can spot it most of the time. But that's my job, right, is to filter that out. I wouldn't say I have any specific skills, it's just that I've seen So many patterns. I can tell on who's doing that.
- [39:55]
A
Right. And then the flip side of that is, you know, you mentioned there's an awful lot of CEOs and leaders out there who, you know, are saying something like, well, I don't run a hospital anymore, I run a tech company. I don't run a car company, I run a tech company.
- [40:09]
B
Yeah.
- [40:10]
A
Is that a good thing? Is it, you know, a mixed thing? Is it sometimes a good thing? What's your reaction to that?
- [40:15]
B
That, well, they're running a tech enabled business. It's. They're really not a tech company at its core and, and they should be more realistic on that. But to be tech first obviously is wonderful. But don't lose track of who your end customer is. Right. So over. So right now we're in this, the hype of AI where everybody's slapping AI, an agent on everything. There will be a point where it just becomes noise and we return back to customer business obsession and, and then AI will just be expected there to provide personalized experiences or forecast what's going to be needed or improve operations. But we're still a period away from that happening. CEOs are talking more and more though to your question about AI and you can read that in the transcripts from the annual meetings. So they know they have to talk about it. Whether or not they're actually doing something is a whole nother question.
- [41:08]
A
Right.
- [41:09]
B
I think with the enterprises, one of the biggest challenges that they have is that to have a great integration with AI, you should have cleaned up data and in most cases their data is already, is a big mess. It's fragmented, there's multiple copies, there's erroneous information. They have multiple customer records. The company's so big, it's, the data is not consolidated, so it's hard to provide a great user customer experience. B2C or B2B as a corporation using AI if the data is already a jump.
- [41:39]
A
Yeah. Well, and you know, the depressing thing is there's no. Unless you have a silver bullet. I have yet to see a silver bullet for cleaning up data. Right. Like it can be done, but it's not like, yeah, we're going to have that clean data for you next week. Right. Like it takes, you know, a fairly large lift to get, you know, an enterprise's data in order.
- [41:58]
B
Yeah. And we'll see AI platforms try to solve that for them and then also fill in missing records with synthetic data. Right. So that starting to happen now.
- [42:07]
A
Yeah. So are you expecting to see like a Leapfrog kind of capability in that space.
- [42:12]
B
I'm operating at startup Silicon Valley, startup speed. So to me the corporations just seem very slow.
- [42:19]
A
Yeah, yeah, makes sense. What's, you know, talking about these CEOs or with them, you know, do you have any specific advice for business leaders or for CEOs who are looking at adopting AI or are trying to get into this, this game and build these capabilities in some capacity?
- [42:39]
B
Yes, obviously they should be using these tools personally and there's been numerous occasions where they're not touching it and they don't understand the power. So they definitely need to activate that. We've referenced how Aaron Levy has been AI first and he's trying the tools and he's encouraging people to try things and that's going to be critical. I see companies doing, I see the largest companies setting up an AI center of excellence. Often it's propped up by one of the largest, you know, the large consulting firms selling million dollar contracts to set up this program management office. So that can sometimes speed things along from a central group and get everybody on the same, literally on the same page within the organization. But then it should be distributed through the organization where each business function is using AI at a rapid pace. Now, if you don't enable a business unit to do that unless you have very strong firewalls, they're going to adopt the open source tools right away. That's why we are seeing growth of enterprise open source tools. And so that's something you need to do. So the CIO needs to move very fast and enable AI within the enterprise or they're going to have a bigger mess on hand and have to roll and wrap everything up into a centralized bet.
- [43:52]
A
So when we talk about the cio, when we talk about this center of excellence and just how we scale this up, One of the pieces that I keep coming back to, and you mentioned it earlier, Jeremiah, is the notion of skills and how we can make sure that we've got the right skills to do this in any organization. Now you mentioned specifically, and it's been fairly topical these days about some of these, you know, multimillion dollars to hundreds of million dollar pay packages for, you know, AI leaders and AI engineers. When you think about the role of skills and how organizations can be better, you know, positioned for this, is it building the skills of existing staff or upskilling them for AI? Is it bringing in AI superstars? Is it a combination of each? Like, how do you structure that problem in your mind?
- [44:47]
B
It's going to vary by company and how attractive those companies are to being forward looking. Obviously the best AI talent is going two directions. One, they join a hyperscaler like Meta is offering those big packages or they start their own company. That's like two routes. They may after that join a traditional tech company, well paid, have a big program to do those things and lots of compute credits. Fourth choice would be joining a traditional enterprise. It's just going to be less attractive. But that's why we're seeing these management consulting firms come into these and offer those solutions. I heard there's one management consulting firm to name. You would know they're doing a billion dollars in AR consulting, AI consulting per year already. It's incredible, right? Because a lot of the knowledge that they're getting you can get from AI unless they it's customer interviews or employee interviews and aggregating that. I'm sure they're doing that of course. So companies are just typically not skilled yet to be at the front of the market. But that's okay. One of the differences with this market versus prior is the no code tools like Lovable or Cursor or Windsurf where business owners, business professionals within the enterprise can start to create their own apps. And so having classes, having courses and curriculum to teach your employees to build apps within a safe sandbox in the enterprise would be one advantage. You're not going to be able to hire the top programmers, so you should teach your existing business leaders and their teams how to use no code to build apps in the enterprise. I think that's an opportunity.
- [46:35]
A
Right. And you know, I love seeing examples of that citizen developer model and what it can unlock for organizations. You did say something that I wanted to come back to though. You talked about the management consultancies and you know, being able to spin up a billion dollar practice. And I, I know I picked up maybe like a note of skepticism in your voice about that model. If you're, if you're a business leader looking to get off the ground in AI, would you be recommending spending, you know, a multimillion dollars with one of the big management consultancies to do that or is there a better or a different way?
- [47:11]
B
It's always best if you can do it yourself. But many of them are not capable to do it themselves or they cannot navigate the internal political turmoil. So you have to hire experts to come in and do it. Those external experts, those management consultants, they've done it at many other companies and so they know how to do that. It's just a very expensive proposition to do change Management?
- [47:32]
A
Yeah.
- [47:32]
B
The ideal sense is that your culture inside of your big company is ready for change management and can have innovation mindset. It's just not built into a company with a hundred thousand employees and 10 levels, 20 levels between the CEO and the frontline workers talking to customers. It's just. It's just not possible in some cases. So you have to bring in those experts to help you. It's just surprising to me that, I mean, you can use AI to actually do a lot of those services now, but big corporations feel comfortable hiring teams to do that and to spend time with them. That physical connection, it just seems opposite to how people operate in Silicon Valley. The AI first mindset, with the startups, you don't hire people first, you use AI first. So that's why you sense that skepticism from me, because I'm seeing something very different on the front edge right here in Silicon Valley.
- [48:26]
A
It makes complete sense to me. And it's funny, if not sad to me about the fact that it's like, if you're trying to do this in your organization, it seems like my sense is, and we have histories in similar industries is like, yeah, you can pull an AI and AI can tell you the answer, but once your organization reaches a certain size, having the answer is not even the hard part. It's like trying to push the boulder up the hill just to get anything done.
- [48:53]
B
And that's where political alignment, Right. Stakeholder management, that stuff that AI can't do. So I think that goes back to your prior question, is like, what skills do I do to train my child? Right. So we talked about leadership, community, empathy. Those are critical in large institutions. That's probably the most important skill. And the skill that AI cannot do, right?
- [49:13]
A
Right. There's a question, Jeremiah, I ask all the guests who talk, who come and talk with me on the show, which is, you know, I'm sure you hear about all sorts of use cases. You know, there's all sorts of, you know, prognoses about what's coming down the pipeline. Is there anything right now kind of in the zeitgeist around tech that you just think is kind of BS or overblown that people should be focusing less on right now?
- [49:42]
B
No, I think the market's focused on the things that matter right now. I don't see a crazy amount of fluff on. There's just too many startups doing the same exact problem. And that that's an issue, but that will go away over time. But I think the industry's focused on business problems like most of the projects that are coming across my desk, they're focused on real problems to solve. The challenge is that there's too many players in the market or they've chosen a market that's too small of a market cap. So like that's an issue. But I don't think there's too much fluff at this time. I would give a warning though is that startups need to follow the principles that I shared before. You need to have an equal amount of focus on your business strategy in addition to your technical strategy. Your technical moat is not enough. It is really not enough. You must focus on your go to market and your business strategies with the network effects and viral effects or frankly, you're not going to survive.
- [50:42]
A
Yeah. And that was a really impactful statement to me. And my sense is that technical mode thing, it's not going to change anytime soon. Right. Like, if anything, it feels like it's going to continue shrinking versus going back.
- [50:53]
B
Should get shorter. Yeah, it'll get thinner.
- [50:55]
A
Yeah. There's one more piece that I've heard you talk about in the past that we haven't talked about today, which is AI influencers. And basically AI as kind of a marketing tool or for human interaction. What's your, what do you see now and what's your outlook there?
- [51:14]
B
Sure. When you mean influencers, do you mean like, like the AI influencers on Instagram? Is that what you mean? Or like AI changing the way we make decisions?
- [51:24]
A
What I mean is more the former or the ability of AI to impact our, you know, call it buying habits or as you know, a marketing channel.
- [51:36]
B
Okay, got it. Yes. All right, so let's break down the logic. And I published about this in 2013. I got picked up in VentureBeat. I was pretty early to publish. Thinking about how the way we make decisions, whether it's B2C or B2B is going to change. So right now the decision process is we think about our problem. We may not know what the problem is. And then we do some Google searches. There's 10 blue links. We go to different websites. The answer is buried at the bottom of the page. We pick up hitchhiker retargeting ads. It sucks. The information we get in one place. Then they want us to contact their sales team. They're going to do through the pricing and then negotiation or we got to shop in different locations. If it's B2C, you have to. You're bounced around to do all these things on the Internet to make a decision and it's not very efficient. Now, with AI agents, the AI agent will be going to those different websites a, collecting that information and bringing it back to you, making suggestions. And then two, what it this means that you don't have to visit those websites or apps. And then two, it'll complete actions for you. It'll, it'll centralize all that information for you in the way that you want, in the medium that you want, text, video or audio. And when you want. So maybe you're listening to this podcast now, but you only want it in written form, but you only want it in 140 characters and you want it at night so it can do that. You don't have to actually listen to the podcast. With an AI agent, it'll do all those things for you. Or if I say I'm going to do a speech on a webinar, you don't have to attend my webinar, you can send your AI agent to attend, get on the notes and bring it back to you. You want to buy a new car? You can send your AI agent to visit all the car sites, give that information back, do pricing and negotiation. If you want to buy something, you can have your AI agent be your representative and purchase things for you.
- [53:22]
A
You.
- [53:22]
B
We bet on Skyfire XYZ to do agent transactions, financial transactions. So this means that the AI agent will be influencing how we make decisions. It will complete the actions and it will recommend products. So search engine optimization doesn't matter as much. Google doesn't matter as much. We will be influenced by AI to make those decisions. And that is a significant change, right? That destroys the media model, that destroys Chipotle, traditional advertising, E commerce changes. Many companies exist by the number of humans visiting their website, cpm. That changes if AI agents are the most dominant entity. In fact, for some consumer sites, if AI agents visiting, do you even need a brand or identity? Right? You just don't need that layer because the data layer and the presentation layer are decoupling and AI agents just pull the data back and reassemble it on a third location. So the way that we're going to make information, the way that we're going to collect information. Excuse me, the way we make decisions is going to be radically changed by agents. They'll be influencing us. B2C and B2B. And I think that's a big change that marketers go to market. Product managers, enterprises, Fortune 500 they're not prepared for.
- [54:37]
A
Right? Which is super interesting and has all sorts of interesting knock on implications. But, but one of them and the one that I've, I've talked about before and I'm, you know, thinking about is if this happens and by the way, I'm pretty sure it's already happening, is that marketers more and more are going to say shit like how do I make sure that I'm influencing the AI agents and the AI tools that are making.
- [55:02]
B
Exactly.
- [55:02]
A
Recommendations. And how long is it before some of these tools are actually earning revenue from generating recommendations from these brands? Right. And suddenly, you know, you're.
- [55:15]
B
Yep, that's right.
- [55:16]
A
Your AI agent is telling you like, you know, you seem, you know, you seem like you could relax and have a Coke right now or something like. And what does, what does that mean in terms of the ecosystem and what does that mean for us in terms of people using these supposedly agnostic tools?
- [55:32]
B
That's a great question. So the biggest question is who does the AI agent serve? Now if you're not paying for it, then you're, it's not serving you. So we're going to see a number of business models premium sponsored free. Amazon offers a free shopping agent called Rufus. Who does it serve? It's free, you know it. There's multiple stakeholders that are served. Right. But if I'm paying for an AI agent and I'm the only one that's paying for it, then it's serving me. Should be added free. We already know that perplexity. They're search search model. So it's not really an agent. But they have already said they want to have sponsored sentences and answers. Right. So that's going to happen. I expect to see E commerce embedded into Claude and GPT in the very near future.
- [56:20]
A
Right.
- [56:20]
B
That should happen. There's lots of prem. There's other things, other after effects that happen to the Goto market team. Like every CRM needs a new field with the associated agent for every customer. B2C or B2B because you'll have multiple agents that are representing the human. Right? That's going to be happening. Every marketer needs to launch sell side agents to converse with the buyer side agent so they have a dialogue and negotiate. All right, we might need to see a new data file, a new data type on your CMS system that just shoots out proprietary data that you want the agents to scrape. In fact, you may sell that data as well. There might be a new business model unlocked by media companies. So there's lots of permutations here on the new business strategies that have to come from that. And that's actually a speech I give to marketers and CMOs nice.
- [57:10]
A
So, so one of the, one of the talking points recently or one of the concerns in this space is that as we move to this world you described where SEO matters less, Google is less relevant. Going to web pages is less relevant. That AI actually in some ways, like actually kills the Internet or the traditional web because that, that whole ecosystem kind of disappears and that's the ecosystem that's been feeding AI this whole time. Is that on your radar? Is that a legitimate concern? And what are some of the issues if we keep going down that road?
- [57:45]
B
And don't forget apps, enterprise apps. Like do you like to go on expense apps to do your expense report? I don't know anybody who does. Do you like going to HR software? Do you like managing your benefits? Like, do you like filling out, do you like going to slack? Do you like to fill out your TPS report? Now you don't have to. Your AI agent will do that for you in one interface. So how many apps do we really need? And you already know my answer. It's equivalent to the number of email accounts you currently manage. That's my general thesis answer to that. So, yes, now we still need websites, we still need new net, new data, we still need net new information. So that will continue. It's just like we need to prepare that the most common visitor will be the representative AI agent, not a human. You need to prepare for zero click searches. When people search on Perplexity or cloud or GPT, all the answers come to them. They don't click on anything. They never need to click on anything ever again. Right. That information has to come to them and be regurgitated in one space. Now, one caveat. Amazing websites that are news and luxury and experiences and sports and social with your family and friends that you actually want to talk to. Yes. You will lean in and you will use those apps. You will still type or talk or video like you still want to do those things. But for all the other stuff that we don't want to do, expense reports and you know, booking a flight, that's very painful actually you send your AI agent to do those things for you.
- [59:14]
A
Right. And it's. Yeah, I mean it. I was laughing to myself as you were talking about, you know, filling out expense reports and some of these, you know, HR apps and yeah, I've got, I'm sure everybody has the same battle scars around those. So it sounds almost Utopian to me. So I, I love that notion. Hey, Jeremiah, just before we go, I wanted to ask if there's any sort of final Thoughts? You wanted to leave our listeners with either human or AI today. Today.
- [59:42]
B
And there are AI listeners, right? They're grabbing the information. They're pre training their models on it. The agents are grabbing information. Right. Like so we definitely have non human listeners today. So you're all welcome. And to the human listeners, what I'm seeing here in the front lines at Silicon Valley is just incredible speed. The key things I'll just repeat is the AI first mindset. Using AI before you do anything else is is the most common thing. Even with your like chores around the house. You don't know how to fix the washer or the dryer. Pull out your AI and turn on the video mode and do a live chat. See if you can troubleshoot. I did that with my microwave yesterday. My vcr finally doesn't blink. 12 thanks to AI. So understanding that I spend 2 hours on average per day listening to AI podcasts to stay on top of the news on average. I enjoy it. I want to learn and it's changing so fast. So I encourage you to figure out how do you consume a lot of information about what's constantly changing in this space? And if you're working at a large corporation, really try to enable a segment of your company to use the no code tools or to be AI first and really lean in and try to adopt these tools as fast as possible. And if you're at a startup and you're building a startup, maybe serving enterprises, please don't forget your business strategies are just as important as your technical strategy. That is one of the most important things that we see companies messing up on.
- [61:12]
A
Amazing. I want to say a big thanks to you Jeremiah for joining. This has been a really, really interesting conversation. I appreciate your insights.
- [61:17]
B
Thank you for having me.