Slideshow with Dave Hayward, Ep 14, Season 2 (#27)
Episode links
Most big companies bought AI licences in 2025 and got almost nothing for it. Ed Ortega’s company went the other way. Everyone there now ships code, including a partner who had never written a line of it.
Subscribe to Bright Objects, our newsletter: Ideas and insights on how to navigate business, marketing, and technology, with a personal lens. Learn lessons from inside ambitious businesses, every week.
https://brightobjects.europa.nz/
Ed Ortega is co-founder and head of technology at Machine & Folk, which builds AI systems for large organisations. He’s also Slideshow’s most frequent guest. He spends his days inside big companies, and his verdict is blunt: most are using a single-digit share of what today’s tools can already do. Some days he feels like a time traveller, arriving from the future to show clients what’s possible.
Dave and Ed work through what AI transformation looks like when it works. Ed’s three-layer pyramid runs from personal productivity, to AI inside the software you already buy, to companies writing their own. He explains why connected AI beats a smarter model, why his team banned AI slop and 20-page documents, and why your new job is to be an AI driver. They also get into marketing spend nobody is accountable for, what’s overhyped and underhyped, and his daughter’s horse jumping game.
The standout moment is a 500-question client security survey. Ed connected Claude to his browser and his company’s policies, let it answer, then had it point him to the answers it was least sure of. “It was the connection that made it powerful, not necessarily its general knowledge.”
Links and things
Slideshow with Dave Hayward podcast links:
Produced by Europa Creative Partners
Chapters
00:00 The models have tapped out
01:15 Travelling from the future to meet enterprise clients
03:00 Why the 2025 AI licence splurge flopped
06:30 Every employee is now a developer
13:30 Multiplayer AI: one brand skill for the whole team
17:00 No AI slop: write for humans, not machines
21:30 Your new job is AI driver
23:30 The AI adoption pyramid
29:15 Software is cheap and SaaS is getting disrupted
32:30 Nobody is accountable for ad spend
39:15 Attention is the new moat
40:00 Overhyped: the AI bubble talk
45:00 Underhyped: solving your own problems
51:15 The models aren’t getting smarter, the harness is
53:00 Letting Claude answer a 500-question security survey
Keywords
AI transformation, AI adoption, enterprise AI, Claude, AI agents, AI slop, AI driver, building your own software, SaaS disruption, marketing ROI, AI hype, data centres, Machine & Folk, Ed Ortega, Slideshow with Dave Hayward
FAQ (for humans and search engines)
Why are so many companies getting little value from AI?
Ed Ortega says many companies bought AI licences in 2025, but adoption stayed low. People weren’t shown how to use the tools, the early tools weren’t great, and most importantly they weren’t connected to anything. “ChatGPT in a void is not super helpful to your day job.”
What does successful AI adoption look like in a business?
Ed describes a three-layer pyramid. The base is personal productivity and skills, such as summarising meetings or learning a new field. The middle is using the AI agents now built into the software you already buy. The top is companies building their own software to solve their own problems.
Can non-technical staff really build software with AI?
At Machine & Folk, everyone now has a GitHub account and has shipped code. The last was a partner who had never written a line of code. Ed set her up in a 15-minute call, and now she updates the company website by telling Claude what she wants.
What is AI slop, and how do you stop it?
AI slop is long, polished-looking output that nobody reads. Ed’s team made an explicit rule against it: use AI to do the thinking, then condense it into a one-page summary written for a human. He says the extra step of humanising the output has lifted the quality of their work.
Are AI models still getting smarter?
Ed thinks the underlying models have mostly levelled off for everyday tasks. The big gains now come from the harness around them: tool use, agents and connections to your email, documents and browser. The more connected these systems are, the more capable they become.
