Episode Transcript
[00:00:01] Speaker A: All right. Hi everyone.
We're back. You're outside. Looks nice.
[00:00:06] Speaker B: Yep. At the new setting, my parents house, enjoying the sun.
[00:00:12] Speaker A: All right, well welcome back Greg and Paul show.
Paul here we break down the Latest in startups, AI SaaS, whatever happens to be debatable on the Internet this week we aim for smart takes.
I know we'll hit on a few scorchers, but we'll always have a lot of dumb ones. You know we love memes way too much.
[00:00:39] Speaker B: Way too much.
Yeah, so we're streaming live on X and LinkedIn every Friday at noon.
Then we post this in video format straight to YouTube. Someday we'll get on to the podcasting platforms as well and I think the plan is we'll do that after we get to the fifth episode. We're almost there. And then we'll launch on Apple, Spotify and Google podcast.
[00:01:03] Speaker A: Well, this is number five, right? That's what we committed to.
[00:01:05] Speaker B: Yeah, that's right, right there.
[00:01:08] Speaker A: Dude, where are you? Looks nice.
[00:01:11] Speaker B: It's my parents house in a little town called Dunville in southern Ontario.
[00:01:17] Speaker A: Ontario, Yep.
[00:01:19] Speaker B: It's actually fairly close to the border, maybe about 45 minute drive to.
[00:01:25] Speaker A: Isn't everything close to the border in Canada?
[00:01:28] Speaker B: Oh yeah, it's true. We all live within like three to four hours of the US border.
[00:01:34] Speaker A: Oh my God. All right, so what's the rundown for today? We've got Soham Prec and I think we'll talk about whether you see him as a pirate or.
[00:01:46] Speaker B: Yeah, pariah.
[00:01:48] Speaker A: We've got some Apple news to talk about. They've done some very weird coding language I know you've got a perspective on.
We want to dive into AI search. A lot happening in the AI search space. There's a Reddit post we're going to look at that has some obvious or non obvious advice for AI search, depending on how you think about it.
Yeah, we've got another Reddit post about launching 39 startups.
Pretty incredible if that's what it takes to have one to hit. But, but we've got the numbers and.
[00:02:21] Speaker B: Of course we'll do the week wild.
[00:02:27] Speaker A: Oh my God.
Did you do anything fun for fourth of July?
[00:02:34] Speaker B: No, I mean we're in Canada so actually we people did set off fireworks, so.
Yeah, so for July 1, which is Canada Day, people set off fireworks. I was here at my parents house July 4th we saw fireworks across the water. So they're right on the water on Lake Erie. You can see a little bit over, over the lake.
[00:03:02] Speaker A: July 1st it's, it's funny how like in Boxing Day, like Canada has all the same holidays. They just are like a couple of.
[00:03:08] Speaker B: Days, a couple of days apart or named different.
[00:03:12] Speaker A: Yeah, they're all basically the same, right? It's really funny.
[00:03:15] Speaker B: Yeah. Instead of like presidency, we call it like Queen's Day.
[00:03:18] Speaker A: Oh my God. That's a real Portland. It was beautiful. And dude, I am like way too old and boring to go to like fire. They. The fireworks started like 10. I was like, 10? Are you kidding me?
[00:03:31] Speaker B: Like, right?
[00:03:32] Speaker A: Like nine. Nine. I'm brushing my teeth at seven.
Like I'm just like, like there's no way. I heard him like in bed and I was like, whatever. Like I've seen fireworks.
I know what they were like.
[00:03:45] Speaker B: The, the cool stuff is drone shows now.
[00:03:48] Speaker A: Oh, dude, they had that too. They were like, there's a drone show. And in Portland they were like framing it as like eco friendly fireworks or something.
[00:03:55] Speaker B: Which. Yeah, which.
Yeah.
[00:03:59] Speaker A: Yeah, dog friendly, right?
[00:04:03] Speaker B: Yeah, go crazy.
I mean, we're getting there. Canada, supposedly they did like a small scale one with like 100 drones or something in. In Toronto for Canada Day.
I mean, some of the other drone shows in asia, it's like 10,000 girls. Yeah, that's crazy.
[00:04:23] Speaker A: I've seen those, like dragon. It's pretty cool.
I like the idea that it's dog friendly. Our little doggy used to get really scared with the, with the fireworks.
[00:04:32] Speaker B: Yeah.
Yeah.
[00:04:35] Speaker A: All right, should we kick this off? Oh, I know we wanted to start with. I had another note in our docket here we want to talk about.
[00:04:41] Speaker B: Nice.
[00:04:42] Speaker A: Is this the peak of the AI boom? Nvidia is now $4 trillion company.
You've got Mark Zuckerberg giving out NBA superstar sized pay packages to AI engineers. And the one anecdote I want to save or I want to share is that I remember the dot com bubble.
And I was on my first business trip ever to San Francisco. I was a kid, I can't remember 1920. I got like an all expense paid trip San Francisco. I thought I was like a rock star and had a hotel that they paid for. And I woke up that morning and I had a copy of USA Today put under my door. That's how long old this is. And it said Cisco, world's largest company. And like, you know, completely naive young kid I looked at and said maybe that's the peak of these bubble things that they always are talking about. And I'll never ever forget that because that was the peak.
The dot com bubble. 99. Maybe it was even 2000 and then all kind of fell apart after that. And so whenever I hear something like Nvidia world's largest. World's largest. Like in the entire world, that's the most valuable company. I think. I think back to my first all expense paid business trip.
[00:06:10] Speaker B: Wow.
So your, your position here is you think the current form of AI, We've, we've reached the top of the bubble because we're making these grand crazy salaries.
Big news. Headline news about Nvidia trillion dollar market cap.
[00:06:31] Speaker A: Yeah. Four trillion, right?
[00:06:33] Speaker B: Yeah, crazy.
[00:06:34] Speaker A: Four trillion. I mean, dude, I ran out of fingers and toes. Like I cannot count 4 trillion. Right.
[00:06:39] Speaker B: Like what does fortunately even mean?
[00:06:42] Speaker A: Crazy.
[00:06:42] Speaker B: That's like a country.
[00:06:45] Speaker A: I mean, I don't know, I know you might feel differently, but I think this, this cycle, there's multiple cycles. Like I say, you remember like an AI boom a couple years ago that maybe was only in the Bay Area. There were a bunch of very interesting AI companies actually that no one talks about now. So we got this new boom, right?
Yeah, I think, yeah.
[00:07:04] Speaker B: So I, I kind of like I have to zoom out in the long term. Right. You can compare it to let's say the other types of improvements in technology.
Let's compare it to electricity for example. Right. I think AI could be the new electricity in terms of just the utility, the many, many types of utilities.
So locally, yes, we could be reaching the top. There could be a lot of consolidation, a lot of standardization. So for example, in the early stages of when electricity became a thing, a household thing, and people are starting to adopt electricity over like burning oil lamps and then buying home appliances, there was a huge boom in terms of just the number of products that people can buy and also literally the different types of plugs that integrates these products into your wall to consume electricity.
Like at one point there was over a hundred different types of plugs on the market.
So basically what that means is if you bought like a lamp and used as your home, you couldn't use it anywhere else because you made your choice. So right now I feel like AI is kind of like the same thing. If you're a company that's building on top of AI and then you integrate with, let's say like ChatGPT, right. OpenAI and suddenly you're building all of these wrappers, all of these products around this core technology because it's not as standardized.
Maybe you're not being able to move it down the road to, I don't know, anthropics model or some new model. Right. Like some Chinese model, open source model. So, like, I think we probably could use some kind of consolidation, like a standardization mvp. Sorry. MCP is sort of kind of a direction towards that, where we're creating this protocol where regardless if you're building on top of OpenAI or anthropics or even Gemini, you can just plug into some AI without worrying about how to interface with bit. So in the long scheme of things, I think in the long, in the, in the longer timeline, I think we are maybe 25% way through to the wide direction.
[00:09:36] Speaker A: All right, all right, I get it, I get it.
Here's why you're wrong.
The definition of is is not that. Is not that it's valuable and useful and going to keep going.
It's that it's massively overvalued. And so I would say, like, I agree with all that. Like, AI is gonna be around. It's probably one of the most revolutionary technologies that's come out since the Internet. I agree with all that. Like, and you have the crypto boom. That didn't really do a whole lot for the world. You had the Metaverse, which, like, I don't think anybody wants anything to do with that. Right. This is real. It's going to keep going. But in my opinion, a bubble is like, the world's largest company is selling shovels. Like, that's not going to last. And then I just don't see it being sustainable to be giving $200 million pay packages to AI engineers. Like, if this is not the peak of the bubble, that would mean that next year we're giving out $300 million pay packages to engineers. And, like, I don't think that that will continue.
I don't have $300 million to bet against that, but I feel pretty strongly that it will not be what's happening next year.
[00:10:50] Speaker B: So it's the peak that I, I somehow. I agree with that.
I can't argue with that. Just like imagining you getting paid to $300 million as an AI researcher at one of these companies.
What?
[00:11:06] Speaker A: That's insane, right?
[00:11:09] Speaker B: Yeah. Like, we're basically saying that these people are once a generation type of, Type of character, type of resource.
[00:11:19] Speaker A: Yeah, Well, I think it's. I think it's a. I think the strategy there is quite. Is actually. I actually think Mark Zuckerberg's in a stupid business person, and the strategy is to, like, take the take. Yeah, no, no, no. Take talent away off the market. I think that's what he's doing.
[00:11:34] Speaker B: Right.
[00:11:35] Speaker A: Like, how would you be incentivized to work really hard at Meta either. Like, sure, maybe you'll provide some feedback and like grab a cup of coffee. But like, right, it's, it's poaching the top people away to hurt competitors is what I think he's doing there.
[00:11:49] Speaker B: Sure, that makes sense. I mean, so he's invested into the parent.
[00:11:54] Speaker A: I don't know if it makes sense, but like, they can afford it.
[00:11:58] Speaker B: They can afford it. Exactly. And I wonder like, what these people are working on, right? Like, what is Meta AI? How does that manifest into the world? Like he's, he's investing to the parent company of Ray Ban.
You and I both wear eyeglasses, so I can totally predict in five years, Facebook comes out with a off the shelf, AR VR AI enabled pair of glasses.
Oh, for sure. And people wear it somehow.
[00:12:34] Speaker A: For sure.
[00:12:34] Speaker B: We don't know. So like there's meta glasses, but if you talk about meta glasses or even wear it out into the public, people are laughing at you as of right now. And the battery sucks.
[00:12:43] Speaker A: So let's, let's move on. We've debated this enough.
Let's go to Soham.
Do you want to show this article?
[00:12:54] Speaker B: Yeah, let's. Let me, let me pull it up and share it.
I, I love this guy.
So for anyone that hasn't heard of this guy, so he. The, the story here is that Suhel on Twitter about two weeks ago, about a week ago, he called out this supposedly engineer in India who works at three to four startups at the same time.
And supposedly he's been going around interviewing YC companies, passing the interview and getting hired, and then either kind of doing the minimum or not showing up to work.
And essentially he's saying he fired this guy after a couple weeks, realized that he is not showing up to work or he's just not producing.
And he's been doing this for lots of time for years.
And a bunch of other founders piled on and says either they've interviewed them or they've hired this guy or literally hired this guy a week ago.
He, that he did really well during the interviews and now this blew up and turned out he's been working at multiple dozens, maybe up to 30 companies.
I, I just blows my mind. I think this guy is a freaking genius, dude.
[00:14:30] Speaker A: How did they not check up on any of his references? They didn't call any of the companies his resume. Like, I don't understand like how he's able to do this to like 5, 10 companies at the same time.
[00:14:42] Speaker B: Right.
[00:14:43] Speaker A: It doesn't make any sense.
[00:14:45] Speaker B: I Think it's just startup culture, right? Like, you move fast, you hire fast, you pass the interview, and then you hire them. So, like, so I. I've actually ran into this.
[00:14:57] Speaker A: You never followed up on, like, a single reference or, like, called somewhere where they worked before just, like, check up or check someone in your network?
[00:15:04] Speaker B: Apparently none of these people did. I mean, what, like, what questions would you even. You just kind of ask them for?
[00:15:12] Speaker A: Did he work there?
[00:15:15] Speaker B: So apparently he did work at a lot of the companies that he mentioned. Right. And he also passed the interview. So he's a good engineer. Just that he's kind of dividing his attention between dozens of companies.
[00:15:30] Speaker A: I don't know what to make of it. I. I didn't understand, like, why this was considered.
[00:15:34] Speaker B: Yeah.
[00:15:35] Speaker A: Why all the founders were, like, upset and why this was like, he's a pariah or whatever. Like, it, like, okay, so it didn't work out. Or for whatever reason, like, this happened with time employees. Like, they don't work out. Like, it's nothing. It's not something new. And, like, so I wonder if, like, was there something else happening? Like, why did this have to get called out and, like, you know, dragged through the.
[00:15:58] Speaker B: The Twitter sphere?
[00:16:00] Speaker A: Like, like, when it's just like, an employee who took on multiple jobs and wasn't able to fulfill his duties in those jobs. Right. That's not cool. But, like, people get hired all the time. It doesn't work out. So I. I don't feel like it's a big deal from that perspective.
[00:16:14] Speaker B: Yeah.
[00:16:16] Speaker A: And, like, I guess, like, I, for me, like, sense of personal responsibility. Like, you're the companies who hired this guy. Like, you didn't call any of the references and do anything to, like, check up on any of this stuff. Like, you have some responsibility. And I felt like they were positioning it as, like, oh, we've poor. Us, like, elite startup founders who've, like, raised money to go and chase their dreams have been hoodwinked by, like, an engineer in India who's got. Who has multiple jobs. Like, I don't know, like. Like just the narrative for me was, like, it did seem, like, necessary to, like, for it to become so public and, like, that the people who, like, hired him, like, hey, maybe they have, like, a role and role to role to play.
[00:16:58] Speaker B: Yeah. Like, call your references. Right? Like, add one more step into your interview process, and then you can catch these things. It's pretty. It's pretty funny.
[00:17:06] Speaker A: Oh, my God.
[00:17:07] Speaker B: So, like, the only thing that I will say on this is there's an entire Subreddit called Overemployed with almost half a million subscribers. That's literally on this topic. So like this is not a new thing. There's literally entire subreddit is dedicated to like how to systematically get hired at multiple full time jobs.
And no one, no one was specific.
[00:17:29] Speaker A: Too, about like what he didn't not do. Right. Was he like. And that was, that's where this, maybe that's where the story gets interesting is like, what was his plan? Like, he's going to take on five jobs. Like not, not do the work. I didn't understand that one either.
[00:17:42] Speaker B: So for from what I understood, he did do some of the work. Like he was, he was pretty good. Like it was. He was legitimately working multiple jobs, full time jobs at the same time.
So I, I think.
Right.
[00:17:57] Speaker A: Like, what is the strategy here? It's like if he got like three jobs, did them all like awesome for him. Like, like crazy talented guy, like wild. But like to get 10 of them or whatever and like not do the work. I'm not sure like what the plan was.
[00:18:13] Speaker B: Right.
[00:18:14] Speaker A: Who is the original from? Sahil. Was that it? Yeah, Sahil, yeah, he was the one. There was a lot of tweets going around and he kind of said the guy's a bit of a psychopath and a compulsive liar. And that seems to be like. I, I like that as an overarching answer to this whole kind of question, like, what was the guy's plan? Like, I think he didn't really have much of a plan. He just was kind of like he found a hack and he started to exploit it and it kind of went awry.
[00:18:41] Speaker B: Correct. Correct. I, I still applauded for trying though.
I, I tap out like 2, 2 dev contracts.
Doing like 20 at the same time is just impossible. Insane.
[00:18:54] Speaker A: It was a. It was and it was entertaining. I liked reading all of it. And Gary was all like, Gary Tan's all like, this is what the value of the Y Combinator network provides stuff which I thought was a bit of an overreach.
[00:19:05] Speaker B: Yeah, he's all over the place.
No way.
[00:19:10] Speaker A: All right.
[00:19:11] Speaker B: No way.
[00:19:11] Speaker A: Gary, let's jump on to your, your, your Mac on here.
[00:19:17] Speaker B: Yes. Let me share this guy.
[00:19:22] Speaker A: So I'll kick this one off a little bit or I'll get started. I'll get it started. So yeah, yeah. Apple released some. I love the way this headline is positioned. Weirdly interesting coding language model.
And then it goes on to compare it to hugging face.
And then it gets real nerdy about all the different types of code bases and models for LLMs versus auto regression versus diffusion models.
Take us through why I thought this one was so interesting.
[00:19:57] Speaker B: Okay, so first of all, anytime that Apple is associated with some kind of model is interesting because they're so freaking quiet on everything AI front compared to Meta, Google, OpenAI, obviously and anthropics. So anytime like they quote unquote release a model, I'm always kind of watching and listening just to see, trying to predict where they're going with their AI strategy. Right. I'm still questioning where they're going with their AI strategy. So the cool thing here is, here's my opinion. They are uniquely positioned to implement a cool different version of AI not in the cloud, which is something called local LLMs. Right. I mean, Apple has billions of devices out there in the wild. They have headphones now, they have their VR headset.
They are one of the only companies maybe beside Google and Android can reach billions of people with their own hardware.
A local LLM basically means that they can deploy these things onto the device, completely detached from the cloud, completely be personalized and secure. And I actually do personally believe that should be a version of the future, how people consume AI.
[00:21:21] Speaker A: Yeah, I agree.
[00:21:23] Speaker B: It's like everybody has a super intelligence personalized, secured to your device in your pocket versus like paying back to the server somewhere.
[00:21:34] Speaker A: I was gonna say, I think there's a lot going on here. Like one thought when we talked about this earlier was like it seemed very unsteeve Jobs, like Steve would leak something really cool and everybody would analyze it, it would make a big splash and we try to figure out what it meant. It's interesting. This one just got like, I don't know, like leaked on a, like an engineering forum or something and you can't tell like is this connected to like a new innovation or not. Like, I really miss the Steve days where the interpretation would be very simple. It's like, oh my God, they're working on some cool local model that's going to go on your phone. And all this mention of like, oh, diffusion models which are used for imaging. It's going to be some cool AI imaging app that like augments your photos and all that. But given the Apple of today, I'm just not confident that's what's going on here.
[00:22:22] Speaker B: I'm not sure I feel exactly the same. Right. I am not sure if Apple has a AI strategy at all. So like, just to kind of like briefly talk about what they're doing here is they have trained on top of a Chinese model called Quantum, which is a autoregressive model and turn it into a diffusal model. First of all, I didn't even know that that's possible, which is really cool.
And the significance here is all of the LLMs that we're using is essentially autoregression model, which is predicting one token after the other one. And the diffusion model is what a lot of image and video generate vendors use.
For example, stable diffusion. It's kind of like how our brain really think about the world. Instead of thinking about things sequentially, we kind of imagine the world in a blurry picture and then you think a little bit more, it becomes clearer and clearer. So any kind of research that moves us towards more of a diffusional model is cool. So I'm glad that Apple's doing it. Right. Like they're uniquely positioned to do this now like you said, they have lots of image data, they have lots of video data. So it's like to them, it's beneficial for them to figure this out.
[00:23:44] Speaker A: Yeah, I mean that's my interpretation of what's going on here.
[00:23:49] Speaker B: Right.
[00:23:50] Speaker A: And you too. Right. Like figuring out how to have like a locally based model on your phone, which sounds cool, but like what would it do?
My guess is that photography and video, that's what the average person is doing with these devices. And so if you could have AI that like improved it, maybe augmented it, like you could put people into some fantasy, generate all kinds of stuff. Like a lot of things that people are doing today online with, you know, any of these like AI deepfake tools. Right. We've used and all that stuff. But just make it really easy accessible on your phone.
[00:24:24] Speaker B: Correct? Correct, Yeah. I mean they also have the. Nobody's talking about it anymore. It's the VR headset.
So like there could be an entire game development, virtual reality thing that could use diffuser models. If you're going to generate the entire world in your headset, you're not going to want it to be generating like line by line. Right. You need that kind of like, oh, it starts off blurry and then it focuses and then you get loaded into this new world.
[00:24:55] Speaker A: So yeah, yeah, I don't know. I don't. We'll see. I mean like I just, I just, I just searched it. So Apple Vision Pro was discontinued, so they don't, they don't offer it anymore as a, as a product. I did try it though and I thought it had a lot of promise. It's just like the hardware for me was like five years out. Like, it was just big. It had like a battery pack. Like, I'm just not going to walk around.
It's so clunky like that. I don't think anyone. Yeah, like, I want it built into my glasses, which I'm actually quite excited about. I think, like, that'll work really well. And then what I like with, like, a lot of things I'm doing with AI recently that I think would be great for glasses is like, a lot of this kind of translation stuff. Like, you know, I go to like the Asian grocer and I'll take a photo with my phone and have it translate the product or the food or whatever. And it tells me, like, how to cook it or eat it used for.
Really cool, dude. I even did this at Whole Foods the other day. I was looking for high polyphenol oil because it's good for my heart. I actually got tested for high cholesterol, which is crazy because I'm pretty healthy.
[00:26:04] Speaker B: But it's like, how's that possible?
[00:26:08] Speaker A: I know, I was like, crazy. So, like, high polyphenol olive oil or oils are a known way to like, reduce it, right? So the whole foods. And I'm like, trying to figure out what has. I take a photo, the entire shelf and ask it. And it tells me that, like, none of them are actually high polyphenol olive oil. And I found one that I thought might be the right option. I took a photo and it was very specific about, like, why that olive oil was the correct one. I was blown away. And I was like, oh, my God, if I had that in my glasses, I could walk in the supermarket and just be like, I need this, I need that. Like, tell me about this. Like, is this healthy? Like, is this good for me? Like, is this vegan? Whatever it is.
So I'm actually quite excited about augmented reality glasses with AI built in.
But no one seemed to really, like, hit the mark in terms of maybe it's like the utility like I just described needs to become well understood. And then people be like, oh, I get why I buy these glasses. It translates things for me, it does things for me. It helps me with this.
If your maps and directions, like walking, I was all built in. You have to leave your phone. It'd be quite useful.
[00:27:15] Speaker B: Yeah, exactly. I don't think we, like, outside of business and from day to day life, we're still kind of just typing into interface. There's so many different ways that we could consume AI. I'm excited about that.
[00:27:27] Speaker A: I think it's going to come eventually. Eventually.
Do you think that.
Because that's supposedly what OpenAI is working on.
[00:27:37] Speaker B: Yeah. There's some kind of wearable, right?
[00:27:42] Speaker A: Or an earpiece. I just, I'm not sure.
[00:27:46] Speaker B: I'm not sure either. Like, it's, it sounds cool, but like.
So here's, here's kind of like the struggle that I see with all of these AI companies. Right. They need more data.
So a lot of their product decisions are not purely kind of thinking for the benefits of the consumer, because there's a percentage of this, whatever thing that they're building has to give them more data to improve their models.
You go down the checklist and it's like the first thing is, okay, yes, it's the benefit of the person who's using it immediately after or very close neck. Neck. How does it get us more data?
You have this computer competing product priority that I think is driving or pulling this product and product idea apart.
I don't think any of these big tech companies has figured it out yet.
[00:28:48] Speaker A: The spoken ones won't actually contribute to the data set.
Like audio, that's what I see.
Right. But like glasses, if I had all that video and visual information, it would.
[00:29:05] Speaker B: It would. Correct.
[00:29:06] Speaker A: Yeah.
So, all right, what do we got? We got some Reddit stuff.
[00:29:14] Speaker B: All right, so we got some Reddit stuff.
[00:29:16] Speaker A: Oh, yeah, I did some doom scrolling so everyone doesn't have to. This one is.
This one. This one's my favorite one. Do you want to bring this up on the screen?
[00:29:24] Speaker B: Yeah.
[00:29:25] Speaker A: It's one of the best Reddit posts I've read in a long time.
This guy posted on indie hackers and he said that he had launched 39 startups until one made millions. This is what I wish I knew. And so the core theme here is something that you and I have talked a lot about, that you got to just test ideas and see what sticks, and that a lot of people just go too hard and too deep into a concept that ultimately the market, for whatever reason, doesn't see value in it. And he proposed a few things in here that I think founders and people who are into building prototypes and MVP should explore.
And one of them was like, he'll build out not just a landing page to test it, but he'll include a checkout and even take people's money to prove that there's demand. And if it doesn't work, he doesn't get enough demand, he'll just refund the money. I thought that was actually quite smart. Like, I would absolutely recommend people do that you absolutely do not want to take people's money. Let me be super clear. Like don't take people's money if you don't plan to deliver them a product. But if you're doing this as some kind of like MVP test and people agree to like all your terms of service and all that, and then for whatever reason you can't deliver it, give it back. And I think that that's really good. It's a really good test of like whether there's demand.
Do you think people, do you think people should take their money and refund it or.
[00:30:57] Speaker B: I, I think that's probably the best way that you can launch a product or begin to even think about building a product.
You should charge people money for it. Everything else is really just fake.
Like waylists, email way lists, people telling you that yeah, we use it without putting their credit card forward. That's all, that's all fake. The best indie hackers I've seen does this right. They, they said, I'll give you a 90% lifetime discount, get it for $50 instead of a thousand dollars when I launch the product so that I can get feedback and see who's actually interested. And if the product fails or if they fail to reach a threshold, like say 10 users in the first month, they just refund the money.
[00:31:49] Speaker A: Yeah, yeah, I believe in this. I think I've actually written a lot to my newsletter recently that like I think the freemium era is over and particularly like indies and early stage startups should like just not even be thinking about freemium anymore. I think the model is broken. I don't think it works. Free trials, I definitely recommend seven day trial, third day trial, maybe an enterprise, you can make up 30, 60, 90 day trial or something, but no more freemium. And then this idea of like, hey, can you get people to pay up? Yeah, and, and determine, yeah, demand and help understand like the price tolerance for things. I think it's, I think it's really smart.
[00:32:32] Speaker B: I think that's super smart.
I think so like with nowadays, especially if you're into hacking or if you're really any kind of like startup founder, your content is the free thing that you're giving away, right? Like your presence and your authority on social media, the value that you bring onto Reddit, onto LinkedIn X, that is the freemium model that you're attracting people to your product or service.
So like by the time that people was clicking to your landing page like they should have already gone or convinced the, the value that you're Bringing or that they would get out of.
[00:33:11] Speaker A: Yeah, I agree.
[00:33:11] Speaker B: Right?
[00:33:13] Speaker A: Yeah, yeah. It fits in my whole, my whole. And so my, my next newsletter I'll go into depth about what I'm calling the MSP Minimum sellable product. And I think that this whole process now needs to be focused on selling because that's what matters. Like MVP is too engineering focused. It's just focused on functionality. And like functional products or engineering functionality is not connected to revenue in any way. It's not, that's not what people are buying. They're not buying engineering functionality. They're buying a solution to a problem.
[00:33:46] Speaker B: Right.
[00:33:47] Speaker A: So you got to frame the problem correctly so that they understand what the solution is and you need to be able to deliver on it from an engineering perspective. But sometimes it's simple, sometimes it's complicated. Right. You just got to find a problem to solve. And I think the MVP process like doesn't take that into consideration.
[00:34:04] Speaker B: It does not take that 100% agree.
A lot of developers or software product builders, they get caught up in the if I build it, it will come mentality.
Maybe this was the case even like five to ten years ago. I would say you could build and launch something because you could be unique enough in your go to market strategy. You could attract enough people to give you a try and if you're lucky enough, you'll build up enough momentum for you to actually launch this into a real company.
I just think the market attention nowadays has shifted so much that for you to just launch something and then hope for enough people will pay attention to whatever you're doing, I think that's, that's not possible anymore. Like you need to, you need to get these, you need to get your customers paying as soon as possible so that they're invested in your success to keep this train going.
[00:35:13] Speaker A: Yeah, yeah, I agree. I, I think like it's changed a lot, particularly in that SaaS based too, like these products are just easier and easier to create.
[00:35:21] Speaker B: Yeah, right. Completely.
[00:35:22] Speaker A: So the idea that like you can spin up an mvp, test it out, like get some feedback, see if people are interested.
I also think there's still an infinite amount of like SaaS products that could be popular and could be successful. I don't agree that the whole space is done and we've invented everything we figured all out. I don't think that that's true at all. I think there's all kinds of new problems, emerging new technologies and new ways to frame it and stuff. And the market just keeps getting bigger too. Right. There's more and more people that just keep coming onto the Internet. And so the idea that like you could build out a niche product. So I've talked a lot about this too. Like, like niche products targeted for different verticals. Right. So just because there's QuickBooks, I met someone who is making QuickBooks for farms and they have very specific strategy around how they market it, what the feature set is. And so if you're a farmer, it's got a lot of value and it's a better alternative than QuickBooks. Right. And I think there's like just tons and tons of ideas like that. And I also think that like, if you're good at it, you create a great service and you solve that problem in a great way, you'll be able to build a bigger business than people think. I think a lot of people look at these and go, oh, the market's not big enough QuickBooks for farmers. How many are there? Like, well, I don't know, there's like a lot of independent or medium sized farms. Then you start thinking like all around the world. Like, it sounds like a really big market to me, 100%.
[00:36:53] Speaker B: There's, I, I love the, there's a quote I'm going to paraphrase here, which is like it's always a race between product builders, software engineers, to build like bigger and better or more foolproof product. And the universe's job is to produce bigger and dumber people.
Yeah. So like I still see a lot.
[00:37:20] Speaker A: Of like founders, they're, I think they're too technically minded. Like they've got a technical solution or, or a, like an, an engineering, they have a feat of engineering that they've achieved. And so their, their perspective is like, how do I monetize that?
[00:37:39] Speaker B: Right, right.
[00:37:40] Speaker A: And what they like to do is like solve engineering challenges. But like what they need to do is solve people challenges. Like they need to like provide solutions that people want to pay for. And sometimes even seen where they're just so excited about solving engineering challenges that they keep engineering stuff that like people don't, don't need or that like the engineering solution convolutes or confuses the situation. So my favorite example is these guys. They were so smart, dude, they were like the smartest team I ever worked with.
And they were like, they had this device and they, they had drivers and compatibility for everything. They were so good at it. They were so proud of it. They're like, they could support to the point like, I think it confused people on like how to use this product because they Were like, yeah, you could integrate it with that too. You could. Like, they had a driver, right? And we were like, why would I need that driver? Like, I don't understand, like, and they just were excited about engineering challenges and they were really good at it. I was like, you guys just pick, like, top five use cases and all you need to support is the drivers for that. Like, you don't need all these other. And then that created another problem too. They have like these libraries of drivers and they had to, like, create a search to find them. There's so many a directory, like, just totally engineering stuff that, like, didn't really need. In fact, like, it. It distracted or detracted from the value prop of the product.
[00:39:02] Speaker B: Right?
Yeah. It's like if you're. If your customer has to think, think to understand what you do that you're.
You're already losing immediately.
[00:39:14] Speaker A: Yeah. They couldn't figure out how to, like, they, they just. They just couldn't.
They couldn't like, focus and pick like one product area and like one. One thing and just focus on that. They just kept like, just engineering, just building stuff and building stuff. And they were great at building things, but, yeah, interesting.
[00:39:31] Speaker B: Do you see a lot of startups who. Launches, get to like, a few customers and then they end up just building more and more stuff? Right.
Is that. Is that like a. Yeah, you see.
[00:39:47] Speaker A: That'S usually, I mean, I think in very general terms, that's usually what happens is like technical founders, right. They like, invent something. They have a solution event they put together.
[00:39:56] Speaker B: Yeah.
[00:39:56] Speaker A: And so they're. They want to just keep building things. They want to keep solving problems and keep like, like adding on and adding more features and functionality or building new things like. Or they get all excited about some whole new area that, like, that, like, doesn't fit with. They have some success, they have some customers and they invent like a new thing. They don't understand that, like, the new thing is not solving the problem of their current customers.
[00:40:21] Speaker B: Correct, correct.
[00:40:22] Speaker A: And sometimes they can't. That one's like a hard one for them. They're like, but I've invented this new cool thing and it's amazing. And like, can you see that it does all this great stuff, like, we get. But that's not stuff that those people need.
[00:40:34] Speaker B: Absolutely. Yeah, exactly. It's like, it's a. It's a weird combination of like, ADHD and a combination of just.
A lot of founders are scared to say no to customers. Especially the customer comes like, okay, if you built this one feature, Then we'll sign on.
That is such a dangerous game to play, especially as you move up the size ladder.
Enterprise customers love to say this to startups. They're saying, well, we can't use you yet, but if you built this, we'll evaluate it in six months.
And it's just.
[00:41:13] Speaker A: This was an interesting one. I worked at one place, we had this advertising product, and then we figured out a way to incorporate some kind of email marketing into it. It's very logical.
[00:41:23] Speaker B: Even.
[00:41:24] Speaker A: I at first was like, it sounds smart. Like, it's all marketing people.
It serves them a need. And what happened was it created a bunch of conflicts and with a lot of our customers because the advertising team and the email marketing team were different teams and they had different KPIs. And if the ad team started using this, like, email tool, it took away from the goals and the KPIs of the email team and create all these conflicts. So even though it was very logical and we thought it was going to be a great product and provide a lot of benefit, it didn't necessarily, like, solve the pain point for our direct customer. People that are running advertising campaigns, they.
[00:42:06] Speaker B: They.
[00:42:06] Speaker A: Even though it was very logical.
[00:42:07] Speaker B: So.
[00:42:07] Speaker A: So, like, it sometimes can be counterintuitive and you need to, like, talk to your customers, right, and figure out, like, are you really solving, solving their problem? And another startup came and came to me recently and I use a tool and I was kind of like, I understood how the tool worked, but it didn't have a lot of confidence that it solved a problem for people. And so I proposed, like, this, like, customer discovery phase. I was like, I think you need to just go and talk to a lot of people, get feedback on, like, what they think of this tool. And, you know, I didn't, I didn't want to give them my perspective. I was like, doesn't matter what I think. What I think matters is, like, hearing from customers, like, do they think this is a good tool? Does it solve their problem? Like, technically it was brilliant. Like, it was really smart. Like, the, the engineering behind it was sophisticated.
But, like, for the user, I was like, it wasn't clear to me, like, what the benefit, what the benefit is, what the benefit.
Yeah.
[00:43:09] Speaker B: Nice.
Shall we?
[00:43:11] Speaker A: All right, should we. Yeah. Move on.
[00:43:13] Speaker B: Yeah.
[00:43:15] Speaker A: We got one more Reddit thread. This is our kind of AI marketing hack of the week.
Okay, so this was cool. So lots of people talking about AI search.
[00:43:32] Speaker B: Oh, this is a good one.
[00:43:33] Speaker A: Yeah, Optimization.
I've seen a lot of people coming to me with tools that do some form of like AI optimization, search monitoring, whatever it is. Right. And so this post that someone did on Reddit was dead on in terms of like providing a lot of feedback and thoughts on how AI mode, let's say AI mode in Google perhaps specifically works and how it's different from SEO. And so they're, let's see there in this post they claim they analyzed 10,000 keywords, 120,000 AIM citations to try to understand the behavior. They had some really interesting results that were quite different from how you'd expect SEO to work. So Rayleigh gives the same answer twice, which I think is obvious if you use an LLM a lot, but important because it's not how search works.
What is this one? 9.2 of URLs are reused across three identical tests. So even the like citations, they're all different and it doesn't necessarily return the same links all the time.
And they said that like their number is 21.2% of keywords had zero reported sources. Right. So it works really different than how Google search does.
[00:45:08] Speaker B: Yeah, it's, it's only going to take over more and more of the traditional search.
So like Google obviously with the AI mode that's kind of already cannibalizing their own search data and they have recently changed the algorithm on how pages are being indexed and also show up in Serpent. This happened I think like July 1st and basically like the third thing that they did is now they have a forum tab on the Google's homepage.
[00:45:43] Speaker A: So yeah.
[00:45:47] Speaker B: Let me see if I can.
Let's go to Reddit. So there is now a forum tab here which groups community websites.
[00:45:57] Speaker A: Yeah, that's a huge deal.
There you go.
[00:46:03] Speaker B: Okay, so that's new whenever that is huge.
[00:46:10] Speaker A: Oh, and that's Reddit and Quora and.
[00:46:14] Speaker B: Basically any kind of communities.
Right. If you're searching up a whatever keyword, it will show you the top communities that have mentioned the keyword in.
So this kind of like tells you everything what Google's next strategy is, which is converting people's traffic from these block type content onto more forum community focused websites.
That's a huge shift and it kind of validates everything that I've been thinking about and planning and talking to my clients about, which is just shifting your content strategy away from writing SEO optimized blog post into just being in community. That's valuable to you?
[00:47:01] Speaker A: Yeah, yeah, 100%. Right. Like it seems like owned and operated websites of a brand are just going to have way less value.
The post actually identifies the top three domains that they saw get cited.
So Indeed was number one, which is interesting. I never go there. But Wikipedia's two, Reddit's three, YouTube's four, and Nerd Wallet was five. So I guess there are. And which maybe Nerdwall falls into like the forum category, perhaps. It's like a content site. It's not a corporate blog.
[00:47:36] Speaker B: They, they have a really good SEO. Like, I've clicked on random search terms that I've landed on and it's like a neural wallet blog post. So they're doing something right. They're hacking something together.
[00:47:52] Speaker A: Yeah, so. So what's interesting too about all this stuff is that, and it was like the first comment right in this thread was, hey, yeah, a lot of the stuff perhaps is obvious. If you do a lot of searches and use this stuff a lot and like, where's the optimization?
Like, it's just, it just we don't understand or we can't necessarily predict the responses.
And so in the old days, or at least the term SEO implies like, that you can optimize it, but it's not clear like that there's any way to optimize or ensure that you get better results or that you appear in the results.
[00:48:39] Speaker B: Yeah. So I think you actually talk about quite a bit about this, which is the concept of like a full funnel marketing.
I, I don't think you can just pick one area and optimize the crap out of it and show ROI anymore that you just can't like say, okay, so SEO is like, beforehand you could just do some SEO.
Six months down the road, you're driving consistent traffic. And now if you implement some kind of like paid strategy, either AdWords, you optimize for clicks and now you have a consistent pipeline. And maybe you can like dabble in social media on the side, get some leads that way, or just build some brand. Right. Like, that's kind of like the typical playbook.
I don't think that.
I think like 90% of people is doing this, or almost 100%, let's say making up numbers here. And because everybody's doing this, the quality of just these three things is like, it's terrible. People are sick of just seeing AI slop when they type in a keyword on Google and then they don't click on Google sponsored AdWord anymore. That's getting super competitive.
And then social media content is also getting super competitive.
[00:50:02] Speaker A: Yeah, no, I agree with all that.
So are you saying that you think like, SEO is dead or generative optimization works?
[00:50:16] Speaker B: I just think that it's part of this. 10 different things now before it's like three things, right? You do SEO, you serve, you buy some AdWord ad on Google and then maybe you have a social media strategy and you have a business like that. That's your entire pipeline. And then like. I know, I know like the stats with user touch point is like seven. That's the number that people keep throwing.
[00:50:41] Speaker A: Around like the right seven touch points before they like book a call.
[00:50:46] Speaker B: I didn't even make a sale, whatever that is. Yeah, I think that number is like more like 15 now.
Like they, they're scrolling on social media. You pop up whether organically or, or paid.
Right. And then they forget about it or clicks on a thing, goes onto your website, bounces, goes on to Google a few days later because they, that problem resurfaces. Saw that your brand is one of the few results on the first page. Maybe there's a Reddit thread, maybe there's a community core question, whatever that is. Right. They don't buy it there. And then, and then a few moments later you hit them with like a targeted ad. Right. Because they, they landed on your homepage whether it's on Reddit, Google or Facebook meta.
So like the touch points could be 10 to 12, a dozen before they even book a call or even think about buying a product.
[00:51:43] Speaker A: Yeah, totally. Yeah. I've seen research on this for like travel. Right, Right. It's like hundreds sometimes like people doing research.
[00:51:51] Speaker B: People are doing more research because the.
[00:51:54] Speaker A: I think you're right. I think, I think like, I think the summary here is that like it's, it is the end of the like Google playbook for marketers that there was a very clear strategy you could employ.
Generally work in a kind of a like a two, a two pronged approach of like SEO semiconductor. And then he can throw some social media sign that seem to kind of like contribute to it. Now we're in a whole new world where the results from Google are, are like there's way less traffic being driven. It's, it's unstable or wildly volatile in terms of like the results.
So marketers need to, for focus on alternatives.
I think like real world events is what's going to come back in a big way. Like I'm doing a bunch of events. We shouldn't talk about that. Doing a bunch of events like next couple of weeks.
[00:52:43] Speaker B: I agree.
[00:52:43] Speaker A: The events are like in huge demand. I'm seeing all kinds of people want to sign up and kind of come and participate, which is nice. Like I always Liked events. I think doing things in the real world is really fun. So I think that's one alternative that is reemerging that we almost kind of over indexed on digital channels because of the pandemic and because they were cheap and because they were new and because they were novel. There's a lot of reasons why. And now I think we're on the other side of that where it's like, okay, the Internet now is this big kind of AI God. And you just kind of ask your questions and it spits back. You just answers and you don't need to do a whole lot more for the average person. They go to the AI God, they're like, tell me, tell me the truth, right? And they just get an answer and they go, thank you. And they just like, go on with their day, right? They're not like us, where they like, spend all this time, like, scrolling through this website, looking at all these different links and comparing things and stuff. And unless they're trying to book their trip to Italy and they're really excited about. Yeah, what, what hotel Cinque Terre. But yeah, I think we're in a whole new. A whole new area. And I think you're right too, that, like, it's fragmented. You're gonna have to focus on a lot of things. Like, you'll need great content. You definitely want to show up in AI Search. You definitely want to do some events. You need some kind of social media. You might need to run some ads, you need to be on Reddit.
You might want to consider some other forums. Like, the playbook has just changed dramatically from like what it was like only a year ago.
[00:54:13] Speaker B: 100%. Totally agree there.
[00:54:17] Speaker A: All right, we got two minutes. You want to do our meme of the week?
[00:54:23] Speaker B: Yes, let's do it.
I. I mean, I think you're. You. You got the, your hand on the meme game every single week.
[00:54:34] Speaker A: Yeah, this one we talked about too. This is the right meme. This is the right one. So this, this photo, I didn't even catch the interview, actually. I just saw this photo and I just thought, like, yeah, it was such. I thought the sunglasses were.
Oh, my God, the glasses are hilarious. Like, what is he. What are they, like meta glasses or something? Like, I don't know what he's wearing.
[00:54:54] Speaker B: I know.
Yeah, the Internet comments is like, this is the glasses he's developing with John. Johnny Ives.
That's.
[00:55:02] Speaker A: Well, yeah, that was, that was my joke, right? Like, no, these are not the glasses I'm developing with Johnny Ives. Right.
[00:55:07] Speaker B: But like, are those things ridiculous? And it's just, it's like mannerism, right? Like the, the reporter, the reporter asked him pretty tough question, which is like, how are you feeling about Mark Zuckerberg carrying away all your best engineers? He's like, oh, I don't think about that. I don't mind.
I don't mind. I. I think, I think, I think the right mission will speak for itself. I don't mind. Like, obviously he might.
[00:55:36] Speaker A: Oh my God. All right. Oh, let's, let's do plug for events. So I'm speaking at bunch of different events over the next couple of weeks. All my newsletter. So I'll be doing how to monetize your climate startup Wednesday here in Seattle next week from 5 to 7 at 90 downtown. You can find links that all over if you do. If you look at any of my social media, it's all over the place. And then Saturday after that, speaking at Venture Mechanics in Bellevue, doing a talk based on my article.
Everything in marketing sucks. What founders can do about it.
And then it's almost like 10 days out. Two weeks from now for Seattle Tech Week on Monday, July 28, I'm doing my talk how to go from zero to a million in ARR for your SaaS or AI startup. That thing is on fire. I've got 200 people RSVP and another 175 signed up. So like go to the other events. If you want to hang out, meet me, chat about SAS and AI, because that one's going to be oversubscribed.
[00:56:49] Speaker B: Amazing.
Nice. And people can find links to.
People can find all of these things.
[00:56:59] Speaker A: Yeah. Beehive and Beehive newsletter. Twitter.
Yeah, it should be easy to find. It's all over LinkedIn too. And do we have one more event we should talk about?
So in October we are. You're gonna be the sponsor. Yeah, you're the sponsor of our SAS and AI mixer at TechCrunch Disrupt in October. October 28th. And we were partnering my buddy Arjun. And it's a small invite only event. We've got a bunch of founders, a bunch of VCs, something that you think you'd like to be a part of, Please let us know and shoot your shot. Pitch us, let us know why you think you'd be an excellent addition to that. Great hair is a good answer.
You're a founder, of course. And if you're an investor, we'll take all three.
Are we good?
All right, dude.
