Code and Capital at 100: PowerPoint, conferences, Harvey AI, and the risks of AI speedDanni is joined by Peter Worn, co-host of World Tech Insider and the writer behind Code and Capital, which just hit its 100th edition. The episode looks at the ideas Peter has returned to most often in his newsletter and why they matter for advice firms, tech vendors, and anyone working in a heavily process-driven industry.We discuss how newsletter writing has become a way to sharpen opinions before speaking, why PowerPoint often obscures more than it clarifies, and why conferences can be both valuable and wildly inefficient. The back half turns to AI hype and AI risk, including Harvey AI, medical transcription errors, and the danger of moving too fast from file note to advice.Key topics:Peter reflects on reaching 100 editions of Code and Capital and why writing has become a useful way to process industry issues before speaking publicly.In this episode, Peter explains why he sees content creation as a labor of love rather than something to do only for output or growth.Peter makes the case that PowerPoint is often overused, especially when a Word document or a better visual format would be clearer.He argues that PowerPoint works best as a storytelling tool when used with strong data, benchmarking, and cleaner design rather than text-heavy slides.We examine the “conference industrial complex” and Peter’s view that the industry spends too much time and money on events, sponsorships, travel, and merchandise.Danny shares a waste estimate for a three-day conference, prompting a broader discussion about landfill, carbon impact, and whether the sector is doing enough to justify the footprint.Peter revisits his Harvey AI commentary and explains why he was sceptical of the company’s valuation, marketing spend, and the way it signals adoption to buyers.We discuss why landing pages that lead with funding totals and user metrics may impress buyers, but also suggest the real purchaser is the CTO or CEO, not the end user.Peter’s “magic mushies” story explores a real-world AI transcription error in healthcare and how a mistaken note contributed to a rejected workers’ comp claim.The episode closes on the broader risk of AI making lazy people lazier, especially in compliance-heavy workflows where a small error can cascade into a bad advice outcome.Peter previews a new presentation, Beware the AI speed trap, built around the need for checks, human review, and consistent firm-wide processes before automating advice generation.Chapters00:00 - Welcome back and Peter’s Code and Capital milestone00:55 - Writing 100 editions and why it became a habit02:07 - Writing to sharpen opinions before speaking03:40 - Internal Slack, shared values, and industry banter05:15 - Why PowerPoint often becomes “death by deck”06:27 - When PowerPoint works and when a Word doc would be better08:33 - The conference industrial complex and who pays09:59 - Too many events, too little time in the office11:35 - Counting the hidden cost of sponsorship and attendance12:02 - Conference waste, landfill, and carbon impact14:08 - Better merch choices and what not to give away15:28 - Revisiting Harvey AI and the valuation hype19:14 - Why leading with funding figures feels odd20:12 - Product usage dashboards as a signal to buyers21:49 - Why Grow still stands out as a prediction22:21 - Super administration, outsourcing, and death benefit delays25:19 - Why squeezed suppliers eventually fail26:29 - Introducing the “magic mushies” AI transcription story27:05 - How a medical note got hallucinated into a false claim29:02 - Why convincing AI outputs can make people stop checking30:11 - AI amplifies both the best and worst traits in users31:07 - The danger of AI transcription feeding straight into advice32:29 - Beware the AI speed trap and the need for QA33:08 - Code and Capital continues
