How Many CTOs

The episode contains a handful of genuinely operational data points - overnight agent runs producing 30 PRs with only 5 usable, code coverage as an agent guardrail, and sub-$400/month token spend - but these are buried in lengthy friend-catching-up meandering, tangents about cats websites and rubber band balls, and broad claims about AI velocity that every B2B tech podcast now recycles.

“I woke up in the morning, I had, like, 30 PRs, uh, five of which were usable. And so I was like, okay, that was a little bit too. I didn't get the right guardrails.”

“Beaker Stack has literally almost 100% code coverage on everything that's there. It's like 99% statement coverage, 95% branch coverage and like 98% function and 99% line coverage.”

The 'classic engineering best practices as the actual unlock for AI velocity' framing is the episode's most defensible contrarian point, distinguishing it slightly from pure vibe-coding hype; however, soul files, agent personas, and the small-team-big-output thesis are now thoroughly circulated ideas, and the six-thinking-hats digression adds no new thinking.

“This isn't about vibe coding. This is about applying classic engineering best practices and driving that velocity by using AI agents.”

“I gave them their soul files, which basically describes their personality, how they work.”

Brad is a practicing fractional CTO who actually shipped a product using the methods he describes, which is more credible than a pure thought leader, but the format is two co-hosts chatting about a personal side project with zero users at time of recording - there is no scale, no revenue, and no external validation of the approach.

“I do still do my fractional CTO work. I'm basically working more than full, full time with multiple different clients.”

“I literally built that in ten days. Ten days of part time coding with me and my four employees.”

The episode does supply concrete named tools (Supabase, Sentry, Stripe, NanoClaw, Resend), specific timelines (May 10 agents start, May 24 Skein coding begins, June 3 recording), code coverage percentages, lines-of-code figures, and a token spend ceiling - more than average - but all metrics are self-reported with no third-party evidence, and there are zero business metrics like revenue, CAC, or user counts.

“my biggest day here, uh, I think was 38,000 lines of code. This is just my contribution, not counting the other four robots.”

“from the moment I forked Beakerstack and said running, it was 45 minutes later. This, the, uh, the initial scan website was up.”

Scott asks useful clarifying questions - distinguishing BeakerStack from Skein, probing agent management workflow, and raising the go-to-market gap - but mostly plays enthusiastic friend rather than skeptical interviewer; he never challenges the lines-of-code metric as a quality proxy, never pushes on the 5-of-30-PRs failure rate, and responds to extraordinary claims with literal 'wow' and 'that's amazing.'

“I would think in 2026 when we were recording this, something like this already exists, doesn't it?”

“How do you manage and operate your workforce?”