
Show Notes
As a business operator in 2026, one of your top responsibilities is to ensure your team knows exactly when AI stops and human judgment starts. Jon and Peter tackle the confusing middle ground between "use AI for everything" and "never let AI touch our work”. Drawing from Peter's engineering background, they introduce the concept of "stamping the drawings". That’s a professional engineering practice where junior staff produces work, but a licensed engineer must review and approve anything load-bearing. Applying that to business, certain outputs demand human oversight. For other outputs, it may not be necessary. Jon’s hard line is, if you're manually transcribing meeting notes instead of using Claude, you're fired. That said, if you're sending AI-generated responses to your biggest customer without review, you're also fired. They discuss why employees are confused when leadership says "use more AI" but punishes them for using it in high-stakes contexts. They argue that, rather than simply enforcing a new set of rules, companies should create frameworks that identify which tasks are procedural (automate freely) versus which are strategic, customer-facing, or load-bearing (human review required). Your competitive advantage isn't some unique AI strategy. It's knowing which decisions only human beings should be making. KEY TOPICS(03:15) The Golden Mean of AI: Why most businesses are either under-utilizing AI or using it so poorly it becomes a liability.(05:50) When AI is Mandatory: Jon’s rule: Why failing to use AI for mechanical tasks like meeting summaries is a "fireable offense".(08:20) Defining "AI Slop": The dangers of letting automated responses run wild in internal communications and high-stakes customer relationships.(11:10) The "Professional Engineer" Analogy: Peter explains the concept of the "Wet Stamp" - why every AI output needs a human signature of accountability.(15:30) System Design & Man-in-the-Loop: How to build a workflow where the buck stops with a human, even when AI does 98% of the work.(19:45) Tool Spotlight: Granola & Claude: Why Jon prefers "invisible" AI tools that augment live conversations without intrusive bots.(24:10) Navigating the AI Backlash: Understanding why the public is souring on AI and how to avoid the "low-level gimmicky" trap.(27:30) Media as Leverage in 2026: Why business owners (from roofers to lawyers) must stop renting attention and start owning their own media channels.(32:15) Permission-Based Marketing: Moving away from interruption and toward building a "tribe of affinity."(37:00) The "Be Weird" Mandate: Why authentic, opinionated content beats "safe" AI-generated posts every time.(47:00) Measuring Success: Why 100 loyal listeners are worth more than 10,000 random views for high-ticket businesses.(50:15) Why there is "bitter triumph" in pushing boundaries. Stay connected for more insights and strategies by following:Jon: @MatznerJon on X and at lazyleverage.beehiiv.com Peter: @pslohmann on X and at peterlohmann.com
More from Lazy Leverage
We mention a lot of books on this show and never write them down. Fixed that.61 books, 39 essays, 88 episodes, free: https://lazy-leverage.saganpassport.com/ Cameron Tope has a rule for his team: use AI all the time and he fires you, use it none of the time and he fires you. Jon has the reason it matters. Underperformers used to give themselves away by producing very little. With AI, they produce more. Cameron is the broker and owner of Emerson Property Management, about 300 single family homes in the Houston area, and he runs it from San Diego. In this one: why Cameron learned the job on the ground before he took it remote. The juggling monkey, and how easy it is to hide in the easy part of a project. Jon's pyramid of leverage, and why he thinks every business owner underhires. Justin Roff-Marsh's idea that you can only maximize one variable. Where TOC falls short. The $500 test Cameron ran with a Sagan hire to rule out a lead channel. And the barnacle hunt, where Cameron pays his team $200 to $300 for the best piece of old process they can find and kill.
We mention a lot of books on this show and never write them down. Fixed that.61 books, 39 essays, 88 episodes, free: https://lazy-leverage.saganpassport.com/ Jon can hand a model 35,000 of his own tweets, 250 newsletters he wrote himself, 400 hours of podcast transcripts, and a swipe file of 5,000 hooks he bought for three dollars. Then he can ask it for 50 LinkedIn posts. If you charge him $5,000 a month to write those posts, that is what you are competing with now, and his argument is that you have to beat it noticeably, not narrowly. Peter has made about twelve short form videos. He started with a DSLR, a teleprompter and scripts written for him, and they were flat enough that Jon told him so. He switched to an iPhone, moved to a different corner of his office, swapped the teleprompter for bullet points on a sheet of paper, and stopped trying to sound like an expert. Now he just does the news. Those ones worked. In this one: the two advantages every business owner has over a full time creator, and why most of them use neither. Why there are fewer people who can make good plumbing content than there are plumbers who can replace a water heater. The barbell Jon thinks every business should run, and the middle most people land in instead. Why Peter sells the sponsorship before he writes the newsletter. And the person already on your payroll you should put in front of a camera full time.
We mention a lot of books on this show and never write them down. Fixed that.61 books, 39 essays, 88 episodes, free: https://lazy-leverage.saganpassport.com/
We mention a lot of books on this show and never write them down. Fixed that.61 books, 39 essays, 88 episodes, free: https://lazy-leverage.saganpassport.com/ Jon's Airtable had one job description copied 135 times, once for every row. Some of it was written by people and some by AI summaries feeding off other fields, because whenever somebody needed a new thing they added a column. Peter runs his Airtable with an iron fist and makes people justify every new field to him personally. Jon lets it sprawl and then goes in with a weed whacker. They land in the same place anyway. Jon tells his one remaining coaching client that he would not put a dime into an application layer right now, and would put all of it into the data underneath. In this one: why venture-backed companies build for scale first while everyone else builds for whatever answers today's question. The five terms that made this click for Jon, including ETL and reverse ETL. Why Peter's market rent lives in three systems that never update each other, and what an owner sees on the portal when they disagree. Why Jon turned down a real recruiting system a year and a half ago and says he would be in trouble now if he had not. Peter's three rules for a software stack. And the three grand a month hire he thinks non-technical companies are all about to make.
We mention a lot of books on this show and never write them down. Fixed that.61 books, 39 essays, 88 episodes, free: https://lazy-leverage.saganpassport.com/ A Formula One pit crew is twenty-two people who work for four seconds a race. Nobody has ever suggested firing most of them, because everybody already knows the crew is there to serve the car. Justin Roff-Marsh's question is why that stops being obvious the second you walk past a customer service desk and catch somebody doing nothing. Justin has been consulting for thirty years and wrote The Machine, the book Peter read years ago and still goes back to. For two decades he has been working inside Theory of Constraints on the part nobody else built, an application of it for a whole company rather than a factory floor. That work is a mini book now, called Ode to Speed, and Peter and Jon have been quoting it on the show since episode 93. His argument is that most people who read The Goal came away with the wrong lesson. In this one: why you pick your constraint rather than hunt for it, and why throwing a dart at one beats getting it right. The two numbers Justin says will run any company, throughput per constraint unit and on-time task completion. The black market for tasks that turns up inside every chaotic business. Why a salesperson takes two years to get good when an aircraft engineer takes two months. And Jon saying out loud for the first time that Sagan is pulling AI back to one centralized team, because a manager with a frontier model will optimize their own department at the whole company's expense. Justin's line for that: a three year old without a chainsaw pulls the cat's tail. Give them one and they dissect it.