The New Way To Build A Startup (8 min)
- Release date: 2026-02-14
- Listen on Spotify: Open episode
- Episode description:
In the AI era, startups aren't winning by hiring faster — they're winning by automating as many internal functions as possible. In this episode of Main Function, Garry breaks down how tiny teams are beating companies 20x their size by building automations into every workflow, from engineering to ops to customer support.
Summary
- 🚀 Claude Feels Like AGI: Recent Claude Code advances make it feel like AGI is here, with Anthropic engineers using multiple instances as AI coworkers for rapid development.
- 💪 Rise of 20x Companies: Tiny teams automate all internal functions with AI, enabling them to outperform giants 20x larger, as coined by GigaML after landing DoorDash.
- 🤖 AI Agents Like Atlas: GigaML’s Atlas handles boilerplate, code, and customer service, letting one FTE manage Fortune 500 volumes by focusing on relationships.
- 📊 Unified AI Interfaces: Legion Health’s source-of-truth tool gives instant patient data access, scaling revenue 4x without net new hires in ops.
- 🔄 Custom Agents Everywhere: PhaseShift builds per-employee AI agents from documented tasks, delaying hires and powering a 12-person team against incumbents.
Insights
How is Anthropic using Claude to mimic human collaboration in product development?
Time: 0:09 – 0:44
Category: AI-Driven Innovation EconomyAnswer: Anthropic engineers manage 3-8 Claude instances for implementing features, fixing bugs, and researching solutions, while humans handle high-level decisions. This internal AI use accelerates development of one of the world’s most sophisticated AI products. (Start at 0:09)
What enables tiny startups to beat incumbents 20x larger?
Time: 0:49 – 3:08
Category: AI in Workforce DisruptionAnswer: 20x companies automate all internal functions with AI—code, support, marketing, sales—making each employee far more productive and delaying hires. GigaML coined the term after closing DoorDash despite being 20x smaller. (Start at 0:49)
How do compound startups evolve into 20x companies with full AI automation?
Time: 1:14 – 2:19
Category: AI-Driven Innovation EconomyAnswer: Building multiple integrated products in parallel combines with AI automating all internals, creating powerful, hard-to-displace product-market fit and extreme leanness. (Start at 1:14)
How does GigaML’s Atlas agent supercharge engineers and customer service?
Time: 3:08 – 4:36
Category: AI-Driven Innovation EconomyAnswer: Atlas handles browsers, policy edits, code, and boilerplate, doubling or tripling engineers’ scope and allowing one FTE to service multiple Fortune 500 clients with millions of calls daily by focusing on relationships. (Start at 3:08)
Can startups combine AI teammates, unified interfaces, and custom agents for unbeatable growth?
Time: 4:36 – 7:37
Category: AI in Workforce DisruptionAnswer: These non-exclusive approaches across engineering, ops, and workflows keep teams tiny, payroll low, and growth record-high, redefining startup building. (Start at 4:36)
How has Legion Health scaled 4x without adding ops headcount?
Time: 4:36 – 6:19
Category: AI in Workforce DisruptionAnswer: A custom AI-integrated interface provides instant access to patient history, scheduling, insurance, and messages, replacing manual searches and enabling a tiny team to handle thousands of patients monthly. (Start at 4:36)
Why does PhaseShift build custom AI agents for every manual task?
Time: 6:19 – 7:21
Category: AI-Driven Innovation EconomyAnswer: The 12-person team documents employee workflows and creates tailored agents, automating design and more to compete with 100+ employee incumbents without hiring entire functions. (Start at 6:19)