Wesley Yu is the Head of Engineering at MetaLab, a product agency working with venture-backed startups at the intersection of design and technology. His route into engineering was non-traditional, starting in media studies and radio production before moving into content marketing at a tech startup in the early 2010s. That shift put him close to the ambition and pace of the Silicon Valley startup world, and eventually led him into a coding bootcamp, learning Ruby on Rails and finding a “maker” mindset through software. He has since spent over a decade building web and mobile products and leading engineering teams.
In this episode, we look at how each major software wave changes user expectations without fully replacing what came before. Wesley frames AI interfaces as arriving into a landscape shaped by mobile responsiveness, sensor permissions, and real time collaborative productivity tools.
Some of the episode highlights include:
Why software “layers” expectations over time, and what mobile taught users to demand from AI products
Designing for latency: keeping users oriented with streaming, status cues, and comprehension time
Making context visible: what an AI agent can see, what tools it can call, and what safeguards exist
Coding agents in practice: when multi agent output helps, and when it harms understanding and conceptual integrity
The biggest interface problem in AI: reducing over trust by exposing sources, uncertainty, assumptions, and verification paths
Q: Where do you see the greatest long-term market value being generated, in AI-enhanced products or AI-enabled products built from scratch?
“I think there are two different ways to look at long-term value. One is where you get power-law winners, where a small number of companies capture an outsized share of value by creating an entirely new category. That is where AI-enabled products really shine. These are products that simply could not exist without AI at the centre of the value proposition. Things like autonomous driving, AI-native education platforms, or AI-first search fall into this bucket, and some of those companies may become category-defining businesses.
But if you zoom out and look at aggregate value across the economy, I think the clear winner is AI-enhanced products. These tools will diffuse across many existing productive sectors where value is already being created and captured. AI will sit in the middle of workflows people already use and amplify them. We already see this with developer tooling, like GitHub Copilot, or creative tools where AI is embedded directly into products like Photoshop or Premiere.
There is still a lot of innovation happening at the model and product level, but the bigger opportunity in the near to medium term is diffusion. People’s habits, skill sets, and organisational structures have not yet caught up with what the technology can do. Because of that, enhancing existing products and processes with AI is where I expect the largest total amount of value to be generated, even if the biggest individual winners may come from AI-enabled products.”
Listen to the Story of Software on:
Apple Podcasts, Spotify, Deezer, & any other podcast platform of your choice. The Story of Software Podcast is produced by Zartis, a software services company. We hope you enjoy listening to this tech podcast and feel free to share any feedback with us:
[email protected]
Ireland