Meet the Slowest Startup Incubator in the World—Pumping Out Billion-dollar Companies
AI & I
Mar 04
Meet the Slowest Startup Incubator in the World—Pumping Out Billion-dollar Companies
Meet the Slowest Startup Incubator in the World—Pumping Out Billion-dollar Companies

AI & I
Mar 04
Sam Gerstenzang and Dan Friedman are redefining startup building—not with AI-first hype, but with patience, domain depth, and human-centered services in resilient industries like medical aesthetics and end-of-life care.
Gerstenzang and Friedman co-lead Boulton and Watt, a deliberately slow incubator that builds companies to $5–10M in revenue before handing them off to professional CEOs. Their ventures—Moxie (empowering nurses to launch medical spas) and Meadow Memorials (a fully remote, real estate–free funeral service now California’s largest)—were conceived and scaled pre-ChatGPT, prioritizing durability over AI-native novelty. Rather than embedding AI into core offerings, they treat it as an operational accelerant: they built 'Matthew Bolton', an AI research agent that synthesizes real customer transcripts and hypothesis trackers to sharpen decision-making—not replace human judgment. Synthetic customer calls failed because they misrepresent authentic psychology; instead, AI excels in early-stage customer discovery, rapidly filtering business ideas and surfacing verbatim evidence. In mature companies, AI yields modest (~10%) efficiency gains, while newer projects see outsized impact—especially when grounded in domain expertise, small-team agility, and willingness-to-pay validation. Their philosophy centers on intellectual honesty, earned insight, and using AI to amplify, not automate, human reasoning.
00:00
00:00
There are two types of companies in the AI era: those building the infrastructure and those applying it to real-world problems
04:28
04:28
Meadow Memorials is the largest funeral service provider in California and operates without physical real estate
04:54
04:54
Moxie launched before ChatGPT and built a self-sustaining business by focusing on market signals and iterative execution
07:31
07:31
Avoiding existential dread by not being directly responsible for the middle years of a single company
13:27
13:27
Med spa core work is not deeply affected by AI, despite early adoption of new medical technologies
18:52
18:52
Mega prompts generate business categories and assess viability, then narrow to a few for human validation
22:19
22:19
Matthew Bolton rereads the point of view, hypothesis tracker, and recent calls to update evidence for hypotheses and guide decision-making
29:35
29:35
Claude can cover a wider breadth but might not produce as interesting or new results as a human
35:00
35:00
Sam Gerstenzang's team trained an AI on his blog posts to reach out to candidates on LinkedIn
37:06
37:06
AI adoption in big companies focuses on internal exemplars, not broad transformation