The ConTechCrew

The episode surfaces a few genuinely interesting ideas - deskilling as a structural risk, context graphs for capturing decision traces, throwaway microcoding - but they are buried under 10+ minutes of weather talk, social media plugs, event promotions, and awards segments. The signal-to-noise ratio is too low for a 37-minute runtime.

“AI is making a lot of things very, very simple to achieve at an 80% margin”

“I feel like it's steroids, man. I feel like it's steroids. You want to look intelligent really fast, just use some AI”

The 'AI as steroids' analogy is memorable and the framing of 'necessary effort' (what should you not cheat on?) is a genuinely useful lens, but virtually every other idea is borrowed wholesale from articles they found (Foundation Capital, Microsoft's Future of Work report) rather than developed from first principles.

“what are the things that require necessary effort to really grow. Like what are the things you should not cheat on?”

“I'm going to build throwaway tech for a long lot this year. You like programs where I'm just going to throw it away”

Jonathan Marsh and Travis Voss are genuine construction-tech practitioners with MEP/electrical sector exposure rather than career podcast guests, but the episode functions as a casual news roundtable between co-hosts rather than a deep practitioner interview, limiting how much hard-won expertise actually surfaces.

“I'm like going to sheet metal shops and I'm like, so, uh, so what's the tribal knowledge inside the sheet metal shop that makes this place work?”

“NFPA 70B, which is about your electrical maintenance plan, used to be a recommendation... And then about three years ago, they're like, nope, it's a requirement”

A handful of concrete specifics appear - NFPA 70B's shift from recommendation to requirement, 129 million building upgrade projects, 100 camps and 1,200 kids for HMSE - but these are all secondhand citations from articles and reports rather than proprietary data or named case studies, and most AI and process claims remain abstract.

“Josh, I don't remember where he cited this from, but he said there's, there's like 129 million like upgrade projects”

“NFPA 70B, which is about your electrical maintenance plan, used to be a recommendation”

The format is a friendly three-way roundtable with no challenging follow-ups, no productive disagreement, and transitions that lean on 'What do you think, Jeff?' The hosts agree with each other throughout, and interesting threads (e.g., where exactly does human decision-making need to stay in the loop?) are never pursued with any pressure.

“What are you thinking? Because I love this super”

“John, any thoughts”