Framework for Building AI Startups from a Product Design Genius

Catalog the most annoying personal tasks that AI could automate.

List every small, repetitive, or emotionally draining task you did in the last week—booking flights, hotels, restaurants, scheduling—and ask whether an AI agent could do it with compute alone.

Why: Pain-driven ideation surfaces high-value, narrow domains where AI can deliver immediate, measurable relief.

"I would pay an AI agent to book me a flight... I'd probably spend a lot of money on all three of those things."

Start with a single clear action, instant feedback, bounded decisions, and zero friction.

Build an agent that masters one action first, add LLM understanding, create tight feedback loops, learn from every interaction, then expand slowly.

Why: Reduces scope, accelerates learning, and prevents the complexity explosion that kills early AI products.

"Start with one action that matters, add LLM understanding, build tight feedback loops, learn from every interaction and expand slowly."

Agent-based startups are the new agencies.

Instead of competing head-on with horizontal platforms, buy or build a niche agency (SaaS, fintech, etc.) and replace its manual labor with AI agents.

Why: Agencies already have paying customers and proven workflows; swapping humans for AI delivers instant margin expansion and a built-in distribution channel.

"Agencies are a really good business right now because it's super manual... agent are the new agencies."

Price at a steep discount to the fully loaded cost of the human labor you replace.

Calculate salary + emotional damage saved, then apply a 50-80 % discount to make the offer a no-brainer.

Why: Anchoring against human cost makes AI pricing feel trivial while still leaving healthy margins.

"I saved you $100,000 worth of labor, $50,000 worth of emotional damage... I'm going to price that at a 75 % discount."

We don’t make services to make money; we make money to make better services.

Decide whether you are mission-driven (Disney/Facebook model) or mercenary; this choice shapes pricing, hiring, and long-term strategy.

Why: Mission-driven companies can raise prices later without churn because customers value the service, not just the price.

"We don't make services to make money, we make money to make better services."

Ship early, remove waitlists gradually, and let early users feel like co-creators.

Introduce the product with a waitlist, let friends and early adopters in, iterate publicly, and celebrate user contributions to foster viral loops.

Why: Community ownership turns users into evangelists and creates organic growth before paid channels are viable.

"We kind of found each other as we were creating this thing... people felt like they helped us build it."

Treat great competitor features as primitives to adopt, not sacred IP to avoid.

Watch competitors, copy brilliant primitives (like wheels), but stay true to your unique vision to avoid getting pulled in a million directions.

Why: Primitives accelerate product quality while vision maintains differentiation.

"These are just primitives... you have to look around and see what all the smart people are doing."

B2B offers tighter, safer outcomes; B2C offers higher dynamic range but lower odds.

B2B lets you interview customers, pre-sell, and build exactly what they’ll pay for; B2C requires catching lightning in a bottle.

Why: Risk tolerance and desired scale should dictate market choice.

"B2C... your odds of failure are much higher... B2B... it's like a tighter range of outcomes."