If I wanted to build $1M+ AI startup in 2025, I'd do this

March 24, 2025

AI Summary

5 min read

If I Wanted to Build $1M+ AI Startup in 2025, I'd Do This

The Startup Ideas Podcast

The guest opens with a warning that cuts through the hype: "It's never been easier to build a million-dollar business, and it's never been easier to lose all your equity value." He points to Cal AI, a calorie-tracking app that uses LLMs to estimate calories from photos. It's a great product—he downloaded it and was impressed—but he predicts there will be a hundred clones in the App Store within two months, and Cal AI will no longer be able to charge what it does. "Those are good for going from a college dorm to being a millionaire," he says, "but they're not going to get you to a deca-millionaire. Your equity value is just toast within a very short period of time because the pricing ultimately goes to zero."

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of The Startup Ideas Podcast

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:00) **The Core Thesis: AI Makes It Easy to Build, Hard to Keep** - The host argues that while AI makes it trivial to build a million-dollar business, it also makes it nearly impossible to sustain that value due to intense competition.
2 (01:46) **Why Most AI "Innovations" Are Just Cost-Cutting** - The guest explains that most AI applications are "sustaining innovations" that lower costs but don't create lasting advantages.
3 (03:11) **The New Moat: Network Effects Over Everything** - The guest argues that traditional moats (small markets, no coding competitors) are dead. The only real moats now are network effects and data advantages.
4 (06:48) **The Playbook: Buy Media, Then Data, Then AI** - The host outlines a three-step strategy: 1) acquire a media business for attention, 2) use that to collect user data, 3) build AI copilots on top of that data for a sustainable business.
5 (12:04) **The "Yachting Website" Example** - The guest clarifies the media-to-data strategy with a concrete example: buy a yachting community site, use AI to build data tools (e.g., yacht pricing/maintenance), then sell services based on that data.
6 (14:22) **The Charlie Munger Lesson & The Coming Flood of AI Content** - The guest reflects on Munger's "investing is hard" wisdom, applying it to the current AI landscape where competition is infinite.
7 (17:39) **The Data Moat is Dead** - A key insight: Frontier models have likely already ingested all public data, including proprietary data from social networks. A "data moat" is no longer defensible.

+ Full timestamped outline available in the app

Show Notes

Join me as I chat with Andrew Wilkinson, Co-Founder of Tiny, as we discuss investment and startup ideas in the current AI landscape. We explore how AI is simultaneously lowering barriers to entry while increasing competition, making it easier to build "million-dollar businesses" but harder to create sustainable long-term ventures. We share specific startup ideas leveraging AI, including secure data integration platforms, automated web design services, and AI-powered lending.

Timestamps:
00:00 - Intro
01:25 - Business building in AI
08:13 - Media Business Opportunities and AI
18:18 - The future of GPT wrappers
21:20 - Automation and AI tools we use
29:14 - Startup Idea 1: MCP and Financial Analysis
34:49 - Startup Idea 2: AI Web Design Agency
37:06 - Startup Idea 3: Message Maxing
43:43 - Startup Idea 4: AI-Based Lending Solutions
49:43 - Andrew's Stealth Startup

Key Points

• AI is making it easier to build businesses but harder to maintain competitive advantages as tools become widely accessible
• Distribution, data advantages, and network effects are becoming the primary moats in the AI era
• Media businesses with high-value niches present opportunities for entrepreneurs to build audiences before layering AI tools
• Secure AI applications (particularly around financial data and personal communications) represent potential sustainable business models

1) The AI Paradox: It's never been EASIER to build a $1M business, but never HARDER to build sustainable value.

Why? Because anyone can "vibe code" your product in a weekend.

Distribution and network effects are the new moats. Tools alone won't cut it anymore.

2) Where are the ACTUAL opportunities?

Andrew & Greg agree: Buy or build MEDIA businesses first, then layer AI on top.

• Focus on High-value niches not mass audiences
• Build community BEFORE tools
• Example: Buy TechCrunch or event series like South by Southwest

3) AI tools they're actually using RIGHT NOW:

• Gumloop - For sales lead intelligence
• Lindy - Email processing & calendar management
• Manus AI - For candidate screening (saved Andrew hours!)
• Vercel V0 - Rebuilt Andrew's pressure washing website in 30 SECONDS

4) Startup Idea 1: "Secure MCP"

Model Context Protocol (MCP) lets LLMs access your data securely.

Andrew built a custom system to analyze his company finances across multiple businesses.

Opportunity: Build the "1Password of MCP" - where security is the selling point.

5) Startup Idea 2: "Message Maxing"

An app that analyzes your text messages to identify:
• Which friends exhibit toxic traits
• How YOU come across in messages
• Who you're neglecting

Perfect for viral TikTok marketing. Could be the next CalAI!

6) Startup Idea 3: AI-powered web design agency

• Use V0 to auto-redesign ugly local business websites
• Reach out with "I already b