Greg Isenberg

WAN 2.2: Alibaba's open-source deepfake video model

Greg recommends WAN 2.2, an open-source video model from Alibaba, for turning a recording of yourself into a chosen character (e.g. a Taylor Swift lookalike). For non-technical users he points to SaaS wrappers like Enhancer.ai and Freepik that skip the GitHub setup.

“So this is using an open source model called Juan 2.2. It's by the Alibaba people. Um, so it is a Chinese app. Yeah. So, you know, beware. You know, for a lot of people using open source technology is kind of difficult. So there are apps like enhancer.ai that allows you just to use their SaaS platform. So you can just pick it and you don't need to like look at a GitHub or there's another one called Freepik, which does the same thing.”

PJ Ace: the AI video ad guy pulling hundreds of millions of views

Greg points to PJ Ace as the number-one AI video ad creator, racking up hundreds of millions of views on AI-generated videos and working with some of the largest companies on the planet, using public-domain figures like Sam Altman and JFK.

“there's this guy by the name of PJ Ace I just had him on my podcast. He's amazing. He's like the number one AI video ad guy, and he's literally getting hundreds of millions of views on his AI videos. And he's, I've seen him, like he'll work with like, you know, the largest companies on the planet. And so he's using, look at this, like he's using Sam Altman here.”

The AI video ad pipeline: script to reference frames to animation

Greg's repeatable process for producing an AI video ad: write the script with ChatGPT or Claude, generate storyboard reference frames with an image model like Enhancer or Freepik, then animate those frames into a cohesive story.

“So first you need to write a script. And if I were to, if you know, what I would do is I would, uh, use ChatGPT, use Claude to come up with a script idea. And we won't have time to go through a tutorial of that today, but then I would create reference frames. So I would use, uh, one of the image models like Enhancer or like Freepik. Which we maybe we can go into later, to create images of what the storyboard of the 20-second or 30-second ad might look like.”

Steal thisWrite the script in ChatGPT/Claude, generate storyboard reference frames in an image model, then animate the frames into a finished ad.

Use an AI browser to find a working discount code at checkout

Greg demos Perplexity's Comet AI browser: at a checkout's promo-code field, he prompts the agent to find a discount code that actually works. Comet reads the screen, searches roughly 10 sources, and lands a working code that saves him $66.

“So what's cool about, uh, Comet is it sees what's on your screen and you can see on the right-hand side, it's, it's looking and it's searching through, you know, in this case, 10 different sources. Um, and it'll hopefully work. Um, and if it doesn't, you can keep just prompting it to get it to work, right? So look, Thrift 10, which is a bad one. Oh, sick. Saved $66 on my 2 jackets.”

Steal thisAt any checkout, ask your AI browser to find and apply a discount code that actually works.

Ask an AI browser to find and play the exact moment in a YouTube video

Greg uses Comet to jump to a precise moment in a YouTube video by prompting it in natural language; the agent downloads the transcript, searches it, and plays the clip, keeping you in flow instead of opening new tabs.

“So the next thing that I've been using Comet for has been, uh, finding exact moments in YouTube videos. So I'll say something like, so I'll open up Assistant on the right-hand side, find and play the exact moment Steve Jobs talks about Apple's intersection of liberal arts and technology. And then Comet's agents go and figure out where that is. And I can— yeah, it basically downloads the transcript and it searches it and look how fast it is.”

Steal thisPrompt your AI browser to find and play the exact YouTube moment instead of scrubbing the timeline yourself.

Use an AI browser to source-hire from LinkedIn and auto-draft outreach

Greg shows Comet finding a specific LinkedIn profile from a natural-language query (someone who worked on Apple AI and is now at Meta), then drafting a recruiting email and offering to send it through your connected Gmail via an agent.

“one of the hardest things I have to do as a founder is trying to hire good talent, right? So you can say, find, um, find and go to the LinkedIn profile of someone who worked on Apple AI and now at Meta. So wow, it'll actually go and do that.”

Steal thisDescribe your ideal hire in plain language to an AI browser, have it find the LinkedIn profile, then draft and send the outreach from your Gmail.

Voice dictation hits 220 wpm vs 45 wpm on a keyboard

Greg cites WhisperFlow voice-to-text as a productivity unlock: typing tops out around 45 words per minute, while speaking gets you to roughly 220 words per minute, making it too slow to go back to a keyboard.

“Like here you can, here you can see like a keyboard, you can only get to 45 words per minute. Flow, you get to 220 words per minute. It's too slow to go back to a keyboard after.”

tryshortcut.ai: a financial analyst-in-a-box AI Excel

Greg recommends tryshortcut.ai, an AI-first version of Excel built by a research lab. You upload financial data and use natural language to get key insights, build DCF analyses, income statements, and more; it also dispatches agents across the web for added context.

“but there's this thing called tryshortcut.ai. It's from this research lab of like ballers, like some of the best researchers on the planet created it. And their first product is to create an AI-first version of Excel. And it's similar to actually Comet in the sense that, you know, on the left it sees your screen, which is like a cloud version of Excel. And on the right-hand side you have prompts and you can just use natural language to tell it what to do.”

Keep sensitive financials local, not in the AI cloud

Greg's stance on uploading sensitive data to AI tools: for truly sensitive numbers, run the analysis locally in Excel rather than the cloud, since you don't know what could happen to data once uploaded. The AI Excel plugin gives you the analyst-in-a-box power without the cloud risk.

“I would feel more comfortable though just doing it locally on, on Excel. So you're not uploading your data to the cloud, 'cause who knows what could happen. Um, so I, I, you know, if it's really sensitive data, I, I do recommend just using Excel. Like, why not? You don't need to have it on the cloud, right?”

Arbitrage: build TikTok meme accounts to funnel traffic to your software

Greg argues the hardest part of vibe-coded software is getting customers, and the current arbitrage is using AI to spin up TikTok meme accounts that you own and then sell your software through them. He cites Realfarm, which automates TikToks to drive traffic and likely does ~$10K MRR or less.

“Well, one way is to own a bunch of accounts like on TikTok that you can send traffic to. So I believe that there's an arbitrage right now to use AI to create these accounts, basically like meme accounts, and then You know, sell your software through them. So there's a, there's a, a piece of software, uh, called Realfarm. This thing must do like $10K a month MRR or less. It's like not many people know about it, but it's a way to automate TikToks to drive traffic to your website.”

Steal thisBuild and own AI-run TikTok meme accounts in a niche, then route that audience to your software product.

AI Apply: auto-apply to thousands of jobs

Greg highlights AI Apply, a tool that auto-applies to thousands of jobs for you and also offers a real-time interview buddy, resume builder, and cover-letter generator. He frames it as an unfair advantage for job seekers in a tough market.

“So there's this thing called AI Apply. Hmm. And if you go to the website, you could, let's say you're looking for a job, you can literally apply to thousands of jobs automatically via AI.”

Sell the disease and the cure: AI Apply now filters AI applications

Greg notes AI Apply lets a million users flood companies with AI-written applications, and when challenged that this hurts employers, the founder said the next product version helps companies filter out AI applications, profiting from both sides of the problem.

“And some guy was like, yeah, but if you're, uh, you know, the company getting these, this sucks. And he responded saying like, the next version of the product is helping companies filter out AI applications.”

Greg spends 10 hours a week testing new AI tools

Greg says he spends about 10 hours every week playing with new AI tools and considers himself proficient or native in roughly 10 of them at any time.

“I, I spend probably 10 hours a week playing with new tools every single week. And I would say I'm proficient in, I'm proficient, like I'm, you know, native in 10 tools, let's say.”

ChatGPT has 800M weekly active users to sell to

At Dev Day, Sam Altman said ChatGPT has 800 million weekly active users. Greg Isenberg frames the new app store as a way to get in front of all of them every week.

“One of the things that, uh, Sam Altman mentioned the other day was he announced that, uh, ChatGPT has 800 million weekly active users. So you can get in front of 800 million people every single week and sell them something.”

ChatGPT has 800M weekly active users to sell to

At Dev Day, Sam Altman said ChatGPT has 800 million weekly active users. Greg Isenberg frames the new app store as a way to get in front of all of them every week.

“One of the things that, uh, Sam Altman mentioned the other day was he announced that, uh, ChatGPT has 800 million weekly active users. So you can get in front of 800 million people every single week and sell them something.”

Facebook almost missed the mobile leap and still became a trillion-dollar company

Greg recalls that Facebook had a bad app and almost failed to make the jump from web to mobile, before becoming a trillion-dollar company. He uses it to argue you'll want an app on the new ChatGPT platform.

“But there was a period of time when Facebook, I'm sure you remember, had a bad app and they almost didn't make the leap to mobile. Now they're like a trillion-dollar company. So I think that there's, I'm not saying it's 100% possible, but it is possible that, you know, this becomes a big thing. And if it becomes a big thing, you're gonna wanna have an app here.”

Why ChatGPT apps beat the old CustomGPTs: contextual discovery

Greg argues CustomGPTs flopped because nobody wants to browse a store and set up apps. The new app store wins because it's contextual: you type a request and the right app surfaces automatically.

“No one wants to download apps. No one wants to go to a GPT store, set stuff up, download apps. What makes this so much better is it's contextual. You just type in something on ChatGPT and yes, you know, you, you might have to mention it, but if you don't have to mention it and you just said, hey, I wanted a logo design and then it pulls up Canva or pulls up another app. The discovery opportunity is huge”

AI Tax Guy: prompt to file your quarterly taxes

Greg pitches an AI Tax Guy app where you prompt 'file my quarterly taxes' and a custom UI shows deductions, autofilled forms, and one-click filing, connecting to Stripe, Plaid, and Google Drive.

“So imagine if you were able to prompt, file my quarterly taxes. It shows a custom UI that shows deductions, autofilled forms, one-click filing. It connects to your Stripe, Plaid, and Google Drive. And someone is gonna build My Tax Guy in a prompt on ChatGPT, and it, it's, it's a huge opportunity.”

Steal thisBuild a ChatGPT app that files quarterly taxes by connecting to Stripe, Plaid, and Google Drive with one-click filing.

BankStatementConverter.com makes ~$40-50K/month solo

Greg estimates BankStatementConverter.com, a one-person site that converts bank statement PDFs, makes $40,000 to $50,000 a month.

“I think it's $40,000 to $50,000 a month.”

The one-person PDF converter with a Berkshire-style site

Greg marvels at the solo operator behind BankStatementConverter.com: a single person running a deliberately ugly, Berkshire Hathaway-looking website with one big button that prints money.

“This dude from bankstatementconverter.com is, it's literally a dude, one person. Look at this website. It looks like the Berkshire Hathaway website, but for converting a PDF. It's awesome. It just, you click a big, big button, convert to PDF.”

AI healthcare concierge that owns the 'find me a doctor' query

Greg pitches an app where you prompt 'find me the best dermatologist near me that takes Blue Cross', connecting to Zocdoc and insurance networks with a map and booking, monetized by affiliate fees or acquisition.

“Find me the best dermatologist near me that takes Blue Cross. Your app connects to Zocdoc insurance networks and then shows an interactive local map and booking options, and then you just make money with affiliate. Like, this is just— or it gets acquired by Zocdoc or some of these big companies. But if you can own this high-demand query, it's worth a lot of money.”

Steal thisBuild a ChatGPT healthcare concierge that owns the 'find a doctor who takes my insurance' query and monetize via affiliate or acquisition.

Ship a ChatGPT app MVP in ~30 days including OpenAI approval

Greg estimates a working ChatGPT app takes 2-3 weeks to build well, plus OpenAI approval time, for roughly 4 weeks / 30 days total to launch.

“I would say realistically 2 to 3 weeks to get something that's working really well. And then you need a probably— well, then you'd have to make sure that you get approved by OpenAI, which, you know, might take some more time, but maybe 4 weeks in total, 30 days.”

AI Grandma: warm, blunt life advice from an 85-year-old persona

Greg pitches AI Grandma, a life-advisor app built around an 85-year-old apple-pie-baking persona who gives old-school, warm, blunt, human advice as an alternative to Huberman-style optimization.

“Sometimes we just need an AI grandma to give us old school, warm, blunt, and human advice. So I think that there's an opportunity to build like a life advisor and you build a brand around like an 85-year-old woman who like bakes apple pies.”

Steal thisBuild a persona-driven AI advice app with a warm, blunt character instead of a generic assistant.

RepairMyCreditScore app for the high-value credit-fix query

Greg pitches a credit-repair app that detects 'how do I fix my credit score' intent, connects to Equifax/Experian, and offers dispute templates and score-simulation sliders, monetized by referral or subscription. He stresses buying a strong domain.

“I think you need to have a really good domain, like repairmycreditscore.com. Um, that's probably taken, but something like that. One of the reasons why bankstatementconverter.com worked is because the domain was really good. So you, you know, there is going to be a bit more of a startup cost here. You might have to spend $10,000, $15,000, $20,000 on a domain.”

Steal thisBuild a credit-repair ChatGPT app on a premium exact-match domain that owns the 'fix my credit score' query.

For SEO, .com is Manhattan and .ai is Williamsburg

Asked whether .ai domains rank as well as .com, Greg says for SEO the .com is Manhattan while .ai is the cool-but-secondary Williamsburg.

“.com is like your, is Manhattan, you know? .ai is like your, it's, it's Williamsburg. You know, it's so cool, but Manhattan's Manhattan.”

Virtual team-building as a service for remote companies

Shaan and Sam spotlight virtual team-bonding startups (Teamraderie, Greg Isenberg's NiceBreak.fun) selling Zoom escape rooms, tastings and improv classes to remote teams. Sam notes NiceBreak charges ~$2,000/month and thinks the space — a toy that becomes a real business — could make tens of millions.

“They're charging a lot of money, which I think is badass. So they charge $2,000 a month Plus, okay, so a grand a month gets you 2 nice breaks per month for, and that's for companies that are 25 to 50. If you're, okay, I understand how this works. Yeah, this could be a hit. I think this is great. I think this could totally be a hit. Can it be a huge VC thing? I don't know, but this could make tens of millions of dollars in September.”

Steal thisProductize remote team-bonding as a recurring subscription (escape rooms, tastings, improv over Zoom) priced in the thousands per month per company.

Selling a literal box of air on eBay for the shipping margin

Greg Isenberg recounts selling an empty box ('a box of air') on eBay for $3 to $7, making money on the shipping markup. He marketed it honestly as a gag gift and even got press for it.

“I basically sold a box of air on eBay for like $3 to $7 and it would make money on shipping. But I would say like, hey, you're buying a box of air and people would think it was funny.”

vidIQ's Rob fired his team, moved to surf in Santa Cruz, and pivoted to a Chrome extension

Greg tells the story of vidIQ founder Rob Sandie, who built a Hootsuite-for-YouTubers, then made a radical bet: he let go of most of his team, moved to Santa Cruz to surf 4-5 hours a day, and pivoted the whole company from a website to a Chrome extension — a move he had to justify to investor Mark Cuban.

“he gets rid of most of his team and moves to Santa Cruz, California, and becomes like, surfs every day, like 4 to 5 hours a day. So now remote is obviously more of a thing, but back then, no. And then decides to completely pivot the business. And because he fundamentally believed, hey, I'm going to forget this website. I'm going to change this to a Chrome extension. Imagine writing that email to your— to Mark Cuban's one of his investors”

The Christian Mingle effect: take one technology and rebrand it per vertical

Greg names a repeatable play: take a generic technology and re-skin it for different audiences. Match.com built basic social-networking tech, then spun out verticals like Christian Mingle where the only real difference is the marketing.

“It's like, right, you know, Match.com basically created this technology, if you can call it technology, um, and I guess it's technology, created some technology, some, you know, basically basic social networking technology, and then focused on a bunch of different verticals, but it's the same under— the only thing that's different is marketing. So the question is, what are other opportunities where you can just take a thing and just brand it for other verticals.”

Steal thisTake a generic product, rebrand it for a specific vertical audience, and compete on marketing rather than tech.

Bundle or unbundle: the only two ways to make money

Greg cites Netscape's Jim Barksdale: there are only two ways to make money, bundling or unbundling. Andrew Parker's classic 'unbundling of Craigslist' post showed the aggregate market cap of the spun-out verticals (Airbnb, StubHub, etc.) far exceeds Craigslist itself.

“It's a famous thing, I think in the '90s, that Jim Barksdale, Marc Andreessen's partner, co-founder of Netscape, basically said there's only two ways to make money. You can either buy bundle or unbundle. And just, you know, there's, I think, a post by Andrew Parker from— he was at Spark Capital, 2012 maybe. He wrote this post about the unbundling of Craigslist.”

Steal thisFind a big aggregator (Craigslist, Reddit), pick one high-energy vertical, and build a dedicated product for just that community.

Anything big enough is going to get unbundled

Greg's thesis: beyond unicorn status there's 'unbundling status.' Any platform large enough (Craigslist, Reddit, Facebook Groups) will inevitably have its vertical communities spun out into standalone businesses.

“My belief is that like there's unicorn status and then there's unbundling status. Anything big enough is gonna get unbundled.”

Discord was the first unbundling of Reddit

Greg argues Discord grew by colonizing Reddit communities: it started in the League of Legends subreddit, recruited mods from that network, then spread horizontally to CS:GO and Dota by building exactly what those communities already needed.

“My belief is that since, you know, 2016-ish, Discord was the first example of the unbundling of Reddit. Then Discord was built on top of Reddit. It started in the League of Legends subreddit. It was a product for that community. Then they went to CS:GO, then they went to Dota. They just basically looked at what they needed. They spread themselves on top of that network. They recruited mods from that network and kind of went horizontally.”

Mine RedditList for what people earn, learn, get paid or get laid

Greg's repeatable process: scrape redditlist.com for fastest-growing subreddits, talk to actual members, and build products for communities looking to 'earn, learn, get paid or get laid.' He marvels nobody is systematically doing this.

“And now, you know, in this COVID land, you know, I think there's so much opportunity to bring people together and people are looking for support communities, either to earn, learn, or get paid or get laid, whatever you want to, however you want to put it. And I don't know why more people aren't just like going on top of Reddit lists. And so that's what I've been doing basically, is basically going in with my team and scraping Reddit lists and speaking to literally people on the subreddit.”

Steal thisScrape redditlist.com for the fastest-growing subreddits, then interview members to find the product they're already hacking together.

The 1% conversion math: 22,000 paying subscribers from a subreddit

Greg's back-of-envelope sizing: 1% of a 2.2M-member subreddit is 22,000 people. If you can get them to pay $5/month for a better product, that's a real business. He believes 1,000+ subreddit-based businesses could each clear $1M/year.

“I mean, you look at that, it's like, okay, 1% of 2.2 million is 22,000. Can you get 22,000 people to pay you $5 a month for better recipes, right? And it's like, the answer is probably. And that's why I actually disagree. I think like there's— I disagree in terms of the amount of businesses that you can create. I actually think that there's probably 1,000 businesses that you could create on top of subreddits and using subreddits as your like idea source.”

Steal thisSize a community business as 1% of subreddit members times a modest monthly price.

The Discord playbook: become one with the subreddit, then build

Greg's full process for unbundling a community: pick a trending subreddit, map the user's week visually with a designer, spend 2-3 weeks asking members open-ended questions about the legacy tools they use, then build the replacement with no-code.

“So, you know, you're at one with the community, you map out the ecosystem, and you should do it like visually, like work with a designer, 'cause I think like actually like seeing it visually is gonna help you. And then you spend some time doing like speaking to people about like, hey, what do you think, you know, in the case of let's say Discord, when Discord was coming out, say, hey, what do you think of TeamSpeak? Do you like TeamSpeak? You know, how often do you use it? Oh, you know, what would you— and ask open-ended questions.”

Steal thisSpend 2-3 weeks asking a community open-ended questions about the legacy tool they hate, then build the replacement.

Sold for a quarter billion, saw nothing: the case for profit sharing

Greg argues common stock often pays employees nothing even on big exits ('we sold for a quarter of a billion dollars... I saw nothing' because investors took it). His fix is old-fashioned profit sharing instead of equity.

“Like, we sold for a quarter of a billion dollars. A quarter of a billion dollars was put in by investors, so I saw nothing. So I think there's actually a huge interesting opportunity to offer— I know it sounds old school, but like good old-fashioned profit sharing.”

Apply slot machines to everything

Greg's distilled insight from giveaways: people's lives are boring and they love slot-machine mechanics, where a small action triggers a chance event. The playbook is to bolt a game of chance onto ordinary purchases and products.

“Number one is that people's lives are probably pretty boring and they wanna make things interesting. Yes. Number two. Number two is that people love slot machines, essentially. Which is you do something, person does something, and something with chance might happen. So it's like, how do you— and I know this is a big gambling me, but like, how do you apply slot machines to everything?”

Steal thisBolt a small game of chance onto an ordinary purchase to boost engagement and order value.

You Probably Need a Haircut hit 150K uniques in 24 hours

Greg Isenberg launched a 'virtual barbershop' (youprobablyneedahaircut.com) during COVID lockdown by seeding it to a few journalists and posting on Product Hunt. It drew roughly 150,000 unique visitors in the first day.

“Like you just seed it with a couple of journalists and throw it up on Product Hunt and before you know it, like, I don't know, we probably had 150,000 uniques in the first 24 hours.”

Greg's teenage affiliate arbitrage: pay $1, earn $5 per install

As an 18-year-old in 2008, Greg Isenberg ran affiliate marketing for eHarmony, Match.com and Zynga, who paid $3-$6 per install. He built landing pages where it cost $1 to acquire a user worth $5 — and admits he invented the auto-playing video pop-under ad.

“Basically, I was doing deals with the eHarmonies, the Match.coms, and Zingas, who were like, they were willing to pay you $3, $4, $5, $6 for every install you generated for them. And back then, there was this arbitrage, I mean there's still arbitrage now, but there was this arbitrage where if you could create a landing page and cost you $1 to get someone to it, to install that game or get that lead to eHarmony, they would pay you $5.”

The investor who passed on Uber's seed round

Greg recounts a very smart NYC friend whom Garrett Camp pitched on Uber's seed round. The friend dismissed it — in Manhattan you just raise your hand and a car comes — missing that Uber was a 10x better product elsewhere.

“Garrett pitched him on the seed round of Uber, and this is in Manhattan. He was just like, "Garrett, I don't know what you're thinking about, man. People aren't going to press a button and a car is going to come. It's just, you raise your hand. Look at me." He raises his hand, a car comes.”

Hold 2-5% of net worth in crypto as a hedge

Greg Isenberg's rule of thumb: everyone should keep 2 to 5% of their net worth in crypto as a hedge. Shaan reveals he's at 9% and Greg at 7.5%, treating overshoot as acceptable.

“I have some crypto. I dabble. I mean, my whole thing is like everyone should have 2 to 5% of their net worth in crypto as a hedge, right?”

Steal thisAllocate 2-5% of net worth to crypto as a hedge — small enough to learn, large enough to matter.

Islands' social metrics: 47 app opens per week, 50% retention

Greg's college group-chat app Islands (sold to WeWork in 2019) had standout engagement: daily users sent 30-50 messages a day, weekly users opened it 47 times a week, the average user invited 2.1 people, and 45-day retention was ~50%.

“Like the daily active users would send between 30 and 50 messages every single day. Weekly active users would open the app 47 times per week. The average user would invite 2.1 people. It was a beautiful product. People loved it. So it had like engagement, it had retention, and retention was 50%, 45 days.”

Metamucil for millennials — D2C subscription fiber brand

Greg pitches rebranding a daily-routine staple like Metamucil as a D2C subscription with a brand millennials connect with, the same way Soylent was conceived as 'Ensure for millennials.' Shaan rates it a solid 9/10.

“I love subscription-based businesses. I love businesses that are repeatable. People who take Metamucil take it every day mostly to get their routine intact.”

Steal thisTake an unsexy daily-routine consumable, rebuild it as a D2C subscription with a brand a younger demographic actually connects with.

The Minimal Viable Press Release

Greg's framework for consumer products: reverse-engineer what would make a good PR story, because you want free customers and word of mouth. He reframes the MVP as the 'minimal viable press release.'

“So, a good framework for thinking about, like, how to think about consumer, like, internet products is, like, what is the thing that's going to get a lot of media attention? Because you want free customers and you also— Yeah, you want free customers, and also word of mouth is the best customer anyways. So like, what's something— like, why don't you just reverse engineer what would be a good PR story? Like, that's just one way to look at like MVP, right? It's like, what's the minimal viable press release?”

Steal thisBefore building, ask what would make a good PR story — reverse-engineer the product from its minimal viable press release.

Take something free, create a new category, charge for it

Greg's idea generator: take something people get for free (water, air) and build a charged-for category around it — the way Evian turned tap water into luxury. He cites the Gumroad founder's ladder: to make $10K/hour, create and lead a new category.

“you know what are cool? Ideas that take something free and then you charge for them. Like, and then I put IG water. Like water bottles. If you think of water bottles, like no one was charging for water until some— someone was like, hey, wouldn't it be cool if like you can transport this water and like we would sell it and then we can have like a luxury brand and like you can have Evian and it's gonna be from France. And then there's this whole new category, like new categories.”

Steal thisList things people take for granted as free, then design a premium category and brand around charging for one of them.

Find the community first, then build the software

Greg's product order of operations: a common mistake is building software then hunting for a community. Instead, find a community with a burning need (Snapchat started in $50K/year LA prep schools) and build the software for them.

“a mistake a lot of people make is they create software and then they find the community versus like finding the community and then building the software.”

Steal thisIdentify a specific community with a burning need before writing any code, then build the software that ladders up to that need.

Chef's Table: a nightly restaurant 'drop' with a live-stream chef

Greg pitches a restaurant 'drop' model — one meal per night that sells out (like NYC's Carbone) — paired with a live stream where the chef explains the dish and wine. Shaan extends it to a Michelin-star IG Live before a $100-$150 high-end delivery drop: QVC for cooking.

“This idea is that like, it's a drop for a restaurant, but you have a live stream with the chef telling you about like if it comes with wine or the meal. It's like, this is how I did it. And it's more of an experience.”

Steal thisTurn a restaurant into a scarcity-driven nightly drop with a live-stream chef walking through the dish — build anticipation, then sell the meal.

$99 lawyer letter on demand

Greg, owed a $5,700 security deposit by his Miami landlord, didn't want to pay a $750/hour lawyer ~$2,000 to send a demand letter. The idea: an on-demand service that sends a scary legal letter to get someone's attention for $99.

“I didn't want to call my lawyer because I was like, at $750 an hour or whatever, it's going to cost like $2,000 for this lawyer letter. So the idea is a $99 lawyer letter on demand. Go scare someone, basically.”

Steal thisProductize the intimidating-but-templated legal demand letter into a flat $99 on-demand service.

Unbundle DoNotPay into a single SEO-driven wedge

Greg pitches taking one DoNotPay use case (e.g. 'get security deposit back Florida'), running Google ads against those exact searches, and using that as a wedge to expand into a broader set of paperwork-killing services.

“It's like the unbundling of Do Not Pay and it's taking this idea and it's like, okay, I'm going to run a bunch of Google ads against like getting security deposit back Florida, getting security deposit back California. And I think there's a big business to be created here, because then that's a, that's a good wedge to sort of expand a bunch of different services.”

Steal thisUnbundle a broad service into one painful, high-intent search query, buy ads against it, then expand outward from that wedge.

NFTs go free-to-mint; selling NFTs dies

Greg predicts NFTs will become ~99% free-to-mint, citing 100 Thieves giving away a free championship collectible that 700,000 people claimed, with creators monetizing via royalties on resale value instead of mint sales.

“And they got 700,000 people to collect. This collectible and it's like a trophy. And then they make money on if there's value, right? So if it grows in value, they get royalties. And I think that's what you're going to see a lot of is like creators and big communities basically doing free-to-mints. Um, even though people seem, you know, people today think that selling NFTs is the future. I don't think so.”