Nov 6, 2024 · 38m · startup-ideas
24 year old making $1m/month from AI shares 3 startup ideas
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Entrepreneur Blake Anderson joins host Greg Isenberg to explain how he generates over $1M monthly building bootstrap mobile AI applications, sharing three validated startup concepts and his blueprint for rapid prototyping, creator-led TikTok distribution, and paywall optimization.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Greg holds 26.2% of the talking time here. How this is scored →
Greg as informed peer 3.5 Guest teaching 4.3 Guest disagreement 1.7 Greg pushing back 2.2
speaking balance: gold is Greg, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 10:35 Blake rejects influencer marketing tools as outright wrong
Blake forcefully dismisses conventional advice about using influencer databases and agencies, labeling it outright wrong and insisting only direct personal outreach works.
Hardest push from Greg ▶ 30:24 Greg challenges competing with Duolingo's new AI featuresGreg directly confronts Blake on the viability of a language learning startup, demanding how anyone could beat Duolingo after they rolled out their own native AI features.
Biggest teaching moment ▶ 24:10 Blake breaks down his nonlinear pricing formulaBlake educates Greg on optimizing paywalls by taking lower per-user LTV to accelerate viral network effects, using Superwall split-testing mechanics.
Greg holds their own ▶ 32:35 Greg cites TikTok Creator Insights language learning surgeGreg demonstrates his own deep domain expertise by citing real-time TikTok Creator Insights data showing massive search spikes for learning Arabic, French, and Spanish.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Greg as informed peer | Guest teaching | Guest disagreement | Greg pushing back | Why |
|---|---|---|---|---|---|---|
| The Untapped Potential of AI-First Mobile Applications | 3 | 4 | 1 | 2 | Greg asks whether apps need an AI hook or if non-AI apps can still succeed on TikTok. Blake explains how GPT vision enabled zero-cost MVPs like UMax and how real-time AI personalization represents a 10x shift in product development. | |
| Startup Idea 1: Career AI All-in-One Platform | 3 | 4 | 1 | 1 | Blake pitches Career AI, an all-in-one suite bundling resume grading, career quizzes, and job prep for $50/month. Greg listens attentively as Blake outlines the concept and basic distribution angles across TikTok and LinkedIn. | |
| Promotional Break: Greg Isenberg's Idea Bank Database | 2 | 5 | 2 | 1 | Following an ad read, Blake explains how he avoids influencer agencies and marketplaces, instead building direct personal relationships and DMing creators' families to break cold starts. Greg affirms his approach while Blake details using ChatGPT and Cursor for MVP coding. | |
| Engineering Feature Hooks and Viral Loops for Career AI | 4 | 4 | 2 | 3 | Greg pushes Blake on how AI journaling differentiates from crowded legacy apps and Peter Levels' failed AI therapist project. Blake defends the concept as an objective, emotional-support second brain rather than a therapist, explaining when AI suffix branding works. | |
| Promotional Break: BoringMarketing.com | 3 | 5 | 2 | 2 | After an ad read, Blake explains his contrarian paywall optimization math favoring user volume over LTV to build organic moat via Superwall. Greg presses him on dark onboarding paywall patterns, which Blake admits are morally questionable but mathematically optimal. | |
| Startup Idea 3: Language AI and Disruption of Duolingo | 6 | 4 | 2 | 4 | Greg challenges Blake on competing against Duolingo after their recent AI features. Blake counters by citing how Cal AI outperformed MyFitnessPal due to agility and incentives, while Greg brings in concrete TikTok Creator Insights search trend data to validate the thesis. |
Statements from this episode (19)
Plug AI scaled to 3M downloads and $200K monthly revenue
“I kind of stumbled into this space a little bit over a year ago with an app called a Riz GPT. It is now named plug AI. We ended up scaling that to a little bit over three million downloads, 200,000 monthly revenue.”
UMax hit 10M downloads and peaked at $700K monthly revenue
“Then built you max that has done over ten million downloads, 500,000 monthly revenue. Now I think it's closer to 400,000, but at our peak, we were around 700,000 in monthly revenue.”
Cal AI reached $1M monthly revenue within six months of launch
“And then most recently Cal AI, which we launched just like a little bit over six months ago and is at about two million downloads and over a million dollars in monthly revenue.”
Viral apps combine AI functionality with e-commerce distribution tactics
“It kind of takes the intersection of new functionality in terms of applications that can be built as a result of AI crossed with novel distribution tactics, largely pioneered by e-commerce people. And so when you take the two of those together, it creates for …”
Anderson built UMax with zero upfront cost using GPT-4 Vision
“With the use of AI specifically GPT vision, I was able to do so with essentially zero upfront development costs with the exception of my own time.”
Fragmented career software leaves room for an all-in-one AI app
“As we all know, there are a lot of new, very popular Fragmented career advancement tools. By fragmented, what I mean is you have a website that you can go on that gives you an analysis of your resume, and it'll give you all the bullet points about how to impro…”
An all-in-one career AI tool could easily charge $50 monthly
“Now I think that a career AI, you charge 50 bucks a month for an all-in-one tool to me feels pretty obvious.”
Using influencer databases and marketing agencies is an ineffective strategy
“Number one, I don't use influencer marketplaces or databases. Number two, I think that there's a lot of messaging on the internet about various tools. So like marketplace and databases or agencies to find influencers for you. That's just like outright wrong. W…”
Anderson used ChatGPT to teach himself how to code
“I, myself, I use ChatGPT to teach me how to code to build RizGPT and UMax.”
Multi-factor rating scorecards turn functional utilities into viral consumer apps
“Now, how do we take something that's functional, right, scanning someone's resume and giving tips on how they can improve it, how do we also make that viral? So, what I would probably do here is a six factor rating system similar to UMACs. I would give them sc…”
An emotional support AI framed as a journal would go viral
“Sort of like an emotional support assistant framed as a journal. Would be pretty easy to get it to go viral as well as I think that there'd be pretty high value add.”
Appending 'AI' to product names works best in stagnant industries
“Appending AI is valuable when you are creating innovation within a relatively stagnant industry, such as Cal AI.”
Founders should avoid 'AI' branding for products requiring perfect accuracy
“If you're building like a medical document service or something where like you need 100% accuracy and any minor errors would be unacceptable. The system just wouldn't work. I would probably not include the term AI as people would be a little bit skeptical or s…”
Anderson recommends Superwall over RevenueCat for mobile app paywalls
“I think that pretty much everybody should be using Superwall. I don't use revenue cap. I only use Superwall.”
Founders should accept lower LTV to maximize total user volume
“I believe that you should optimize for like number of users raised to the N where N is greater than one multiplied by profit per user. I will take at least at a just pure conversion basis. I will accept a lower LTV on a per download basis in exchange for more …”
Placing paywalls at the end of onboarding outperforms all other strategies
“The most standard practice in the industry right now is to take people through some onboarding flow and then hit them with a paywall at the end. It's just like, you make so much more doing that than pretty much any other paywall strategy.”
MyFitnessPal engineers lack equity incentives and care less about the product
“The engineers at my fitness pal, like my fitness pal to begin with has been like traded back and forth, sold so many times that the people that are doing the groundwork don't like, probably don't have that much equity, you know? And so the engineers don't care…”
Duolingo's dominance discourages competition, creating opportunities for new AI apps
“For the same reason that Duolingo is so big, there's so, so few people trying, which just makes it so much easier to succeed when applying these sort of new wave tactics.”
Anderson uses highly curated burner accounts to research consumer behavior
“When I was building you max, I had a burner account where it was like only looks maxing content or anytime I build a new application, I create a burner account and I curate my entire feed around that and like put myself in the mind of someone who's consuming t…”