“I’m coming around to the view that running human-written code – without it being AI-audited – is going to be considered the reckless thing.”

In this episode, Martin Alderson – cofounder of Catchmetrics, engineering leader with two decades shipping enterprise software, and author of a Top 100 Hacker News blog on the intersection of software engineering and economics – cuts through the AI noise with the rare combination of someone who actually builds with agents every day and can read what that does to markets. The thesis that runs through the whole conversation: most companies think their AI problem is a tooling decision. It’s not. It’s a mindset problem, and the wrong one is quietly existential.

What You’ll Discover:

[01:29] The $285 Billion Markdown File

→ Why Wall Street wiped out a quarter-trillion in SaaS value over 13 markdown files – and the three distinct ways every SaaS company is now exposed.

[05:38] The Figma Trap

→ How SaaS companies ended up paying their single biggest competitor every time a customer uses their product – and how many others are in the same corner.

[12:33] The 180 Nobody Saw Coming

→ Why the “AI writes insecure code” fear is flipping on its head – and the case that human-written code without an AI audit is about to be the reckless choice.

[15:55] Why Non-Technical People Out-Build Engineers

→ The reason marketers and PMs are shipping products in a weekend – and the human psychology that lets them iterate without the guilt that kills good ideas.

[23:45] Why a Copilot License Can Kill Your Company

→ The bifurcation of intelligence: how a “safe” enterprise tool convinces leaders AI is mediocre, right as they’re about to be out-executed.

[28:33] Five People vs. a Thousand-Person Company

→ Are small teams with agents winning a window that closes when the frontier labs steamroll them – or is this the new permanent state?

[33:14] The One Question Every CEO Needs to Answer

→ The deceptively simple question Martin puts to leaders – and why most boards can’t even answer it.

[35:50] The Oligopoly Scenario

→ Why open-weights models are quietly closing up, and the future Martin is genuinely worried about: three or four giants extracting rent with little incentive to innovate.

Key Takeaways:

• The barrier to building complex software has collapsed – but the threat to incumbents isn’t customers rebuilding a CRM, it’s a five-person team out-executing them on cost and speed.
• Don’t measure AI adoption by Copilot licenses issued or people trained. Ask how many tokens your organization is actually consuming – and whether you can even answer that at a board level.
• The real strategic question isn’t “how do we add AI to what we do?” It’s “if we reimagined the whole business with ten great people and agents, what would it look like – and what’s blocking us from getting there?”

About Martin Alderson:

Martin Alderson is cofounder of Catchmetrics and the writer behind martinalderson.com, a Top 100 Hacker News blog covering the AI transformation through the dual lens of software engineering and economics – read and recommended by people including OpenAI co-founder Andrej Karpathy. He’s spent two decades shipping enterprise software and writes some of the sharpest analysis anywhere on where AI, building, and markets actually collide.