
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.
Live, messy, human content is becoming more valuable as AI sanitizes everything else — Jonathan Courtney points to Twitch-style creators like Jeff Gerstmann and the tech show TBPN/TVPN, arguing that unscripted business streams with even 1,000–2,000 loyal viewers can support €400,000–€500,000 businesses through events, merch, and Patreon-style monetization.
The next big app wave is 'action apps' that do work for you instead of asking you to click around — Greg frames AI-first mobile products as agent-driven replacements for tools like Gmail or Superhuman, where software clears inboxes, books calendars, and files expenses rather than just showing you dashboards.
Loneliness is both a social crisis and a startup category hiding in plain sight — With nearly 22% of Americans reportedly having no close friends, they highlight businesses like 222 and Fabric, plus hyper-niche communities like the 13,898-member 'Dads of Marathon' Discord, as proof people will pay for connection online and IRL.
Older adults are massively underserved, and founders are still irrationally fixated on Gen Z — Greg’s 'elder tech' thesis is simple: there are 70+ million boomers in the U.S., they have money, they have real pain points around hearing, mobility, memory, and social life, and most startups still market like everyone is 24.
'AI employees' will win when they’re sold as narrow, obvious job replacements, not abstract automation — The clearest path, they argue, is to verticalize hard: pick one role like junior YouTube producer or ops assistant, map the 50 jobs-to-be-done, and offer a digital worker that does a subset at one-tenth the cost.
The best startup ideas often come from pairing a real niche with spending power, not chasing the flashiest product — Courtney’s own retreats sold far above category norms because they targeted entrepreneurs, leading to Greg’s line: 'date the product, marry the niche.'
Jonathan Courtney opens by betting hard on live, unscripted creator content. He points to Twitch streamers like Jeff Gerstmann and gaming-style formats as the model, then connects it to business media: as AI makes everything smoother and more generic, audiences will crave the messy, human, can-go-wrong energy of live shows. He says even his relatively small audience of 1,000–2,000 weekly viewers can support a €400,000–€500,000 business without ads.
Greg’s first big category is what he calls action apps: AI-first software that takes action on your behalf instead of waiting for you to click through menus. His example is email — not Gmail with AI bolted on, but an inbox that learns your style, answers 80% of messages automatically, and only surfaces the one or two truly important threads. They compare it to the mobile shift, arguing incumbents may struggle to become 'agent-first' the way old web companies struggled to become mobile-first.
Jonathan then zooms in on loneliness, calling it both a huge startup opportunity and something that 'needs to happen for the good of humanity.' He shares a five-day off-grid retreat he ran for $90,000, with no phones, no internet, and no tech — just people doing art and hanging out — and says it broke even but proved the appetite is real. The emotional proof point is his obsession with 'Dads of Marathon,' a 13,898-member Discord for dads and casual players of the game Marathon, which he says has become his first real online community in years.
Greg backs that up with stats: nearly 22% of Americans have no close friends. He highlights 222, which matches people for dinners, cocktail bars, salsa nights, and basketball through a personality test, and Fabric, which runs 75-plus gatherings a month in New York and Chicago for around 500 members. The recurring theme is simple: niche communities plus memberships equals recurring revenue, and the more specific the identity — dads, hobbyists, a city, a game — the stronger the pull.
Greg’s 'elder tech' idea is a direct shot at startup groupthink. While everyone builds for Gen Z and millennials, he argues there are 70-plus million boomers in the U.S. with real unmet needs and purchasing power, especially around memory, hearing, mobility, vision, and social life. Jonathan adds a practical founder lesson from his own company: when they ran broad Facebook ads, their best customers turned out to be 45-plus, not twentysomethings, because older customers actually had money and urgency.
Jonathan’s next pick is adults learning hobbies for joy — painting, gardening, music, woodworking — especially for people aged roughly 35 to 65 who are emerging from years of nonstop work or parenting. He says his own painting retreat sold out almost instantly, and Greg riffs on that with the example of comedian Tyler Labine’s Creative Club-style workshops where people make paper lamps, drink tea, and go on first dates without staring at phones. Their point: anti-screen, anti-optimization experiences are becoming premium.
When they get to AI employees, the conversation gets more tactical. Jonathan says the real opening is not replacing senior creative people, but automating junior, repetitive tasks — things like chaptering podcast videos, generating clip candidates, or handling operational grunt work. Out of that comes a concrete startup concept: 'Juniors,' a company that sells AI workers for junior-level jobs while explicitly promising not to replace human creativity.
The last stretch ties several ideas together. Jonathan wants more verticalized health startups after describing his own GERD struggle and the absurdity of exporting blood tests into Claude just to make sense of them; Greg cites Zoe and says the real opportunity is building disease- or condition-specific versions of personalized nutrition. They close with pet health and AI-native media, but the big meta-point is Greg’s best line of the episode: 'date the product, marry the niche' — because the founders who win will pick an underserved audience with money, pain, and a reason to care deeply.
Share
Keep Reading
The Weekly Echo. The inbox-shaped summary of what mattered.
New editorials announced here.

Playbook
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.