On Friday Denis Čahuk and I had a great discussion about AI, mentoring junior engineers, and what software development is really about.
I can’t remember if I first encountered Denis through his Substack Crafting Tech Teams or on one of his LinkedIn Events, but it was a bit of a full-circle moment to be broadcasting a conversation with him on both platforms!
I have learned a lot from him when it comes to the craft of software and the importance of shaping human teams to achieve better outcomes, so I’m grateful for the chance to discuss these topics. Hopefully, you find it useful and interesting too!
(And apologies for the internet issues halfway through - as flies to wanton boys are we to the WiFi gods).
AI Summary
The biggest challenge with AI isn’t technical—it’s economic and organisational. Unlike previous technology shifts, AI creates unusual incentives where vendors selling AI tools are the loudest voices telling you that you’re doomed without them. This conversation explores three critical tensions: the gap between using AI for internal productivity versus embedding it in products, the surprisingly personal and opaque nature of working with AI compared to the transparent culture of web development, and the danger of junior engineers treating commoditised problems as novel challenges. The central insight is that startups should focus their smartest people on their unique core product while using AI (or any tool) to accelerate the commodity layers—but only after understanding which is which. The real competitive advantage isn’t adopting AI first; it’s knowing where your actual differentiation lives and protecting your ability to iterate on that as quickly as possible.
Chapters
(00:01:42) - Introductions and Setting the Stage
(00:03:16) - Why AI Represents the Biggest Disruption in a Decade
(00:04:47) - The Anxiety Around AI Adoption: Consumer vs Developer Use Cases
(00:06:11) - Economics and Vendor Incentives Driving AI Hype
(00:07:01) - The Dark, Personal Nature of AI Interactions
(00:09:38) - Why AI Tools Feel Different Than Traditional Software
(00:15:36) - Learning AI Through Usage: The Bootstrap Problem
(00:24:12) - Where AI Actually Helps: From Generic to Specific Solutions
(00:32:45) - AI as a Productivity Multiplier for Engineers
(00:42:18) - The Real Question: Build vs Buy in the AI Era
(00:51:33) - Junior Engineers, AI Tools, and the Risk of Cargo Culting
(00:58:14) - Organizational Patterns: Where to Place AI Capabilities
(01:04:15) - Accountability and the Human in the Loop
(01:06:34) - Commoditized vs Custom: Mapping Your Software Architecture
(01:08:28) - Wrap Up and Where to Find Alex
Key Moments
The Biggest Technological Shift in a Decade (00:03:22) - “This is the biggest disruption, biggest change I’ve witnessed. I’ve been in startups for about 10 or 11 years... So I think it is a big deal, but there’s also a lot of BS around it as well or there’s a lot of hype.”
The Vendor Hype Machine (00:06:00) - “It is vendors who are offering AI. They are telling you that you’re doomed if you’re not buying their product, which is... it’s difficult.”
AI’s Opaque Nature Creates Shame (00:08:11) - “I’ve often had some really interesting back and forth with an AI chatbot and got to something and I’m like, oh, this is really interesting. And a bit of me goes, oh, I’d love to share my whole chat... but then some of the questions I’ve asked or the way I’ve asked it... I might look like a bit of an idiot.”
The Economics of Engineering Are Fundamental (00:06:15) - “One of the things I always bang on about is how important economics is in engineering and it’s so fundamental to how software gets built. And I believe also how to build software well, if you’re not factoring in economic considerations.”
Focus on Your Unique Core (01:06:24) - “Figuring out, getting to that point of what is the unique thing here, the insight into the market we’re generating, or the thing that we’re building that no one else can. And that’s what I help people get to as quickly as possible.”
Commodity vs Custom Software Blindness (01:07:02) - “To a non-technical co-founder, the line would be, oh, everything is custom software. Nobody has ever done this before because our logo’s on it. It’s unique, right? But no, it’s actually like 80% of what you need is a solved problem.”
Sharing is caring!
If you enjoyed this conversation, please consider sharing it with two friends who might find it useful.
