When I first started my market research business: • Couldn't get clients • Hated writing reports • Struggled to get people to act on my insights 5 years on, our client portfolio is 🔥 and we've 10x'd revenue This journey taught me 5 lessons that will help you too:

This means we need to start voting for AI pilled candidates.

Politics of AI: this article comes out a day after Anthropic revealed enemy states using Claude to fight US and a week before Xi makes a US visit.

Today in Omarchy: "Build a tool that will share the clipboard text + images between linux and my mac." A few minutes later I'm here posting about how it just works.

Ryan Holliday was so cool when he wrote that book about Hulk Hogan and Peter Thiel vs Gawker.

Bernie would be much happier working in the EU.

As agents write more of my code and do more of my research work, I find my codebase moving from the right to the left. Much simpler, more effective, easier, faster to load into context and use.

10 years ago I heard about Jobs to be Done (JTBD) and made it the focus of my career. Here are my my biggest mistakes about JTBD: 1. I was dismissive of the work of Tony Ulwick. His work is so practical for anyone who is building something and wants to make it valuable for customers. I simply do not think you can make a great product without applying his JTBD thinking and/or ODI. "Was dismissive" but no longer. 2. I was drawn in by listening to Bobby Moesta interviewing someone about a Porsche they had purchased. I thought whatever Bob was doing -- I want to do that. But Bob is great at what he does because of his unique experiences and skills (and constraints like being dyslexic). 3. JTBD (Moesta) is like a portal into the customer's life and it allows you to see with high clarity how they desire and decide things and looking at the world this way opens up beautiful patterns. I can't imagine trying to create something new without looking through his lens. BUT 4. That JTBD interview is powerful for everyone in the room. But so much of what you learn/experience you gain by having that conversation is tacit. It upgrades your OS, but is really hard to help others get as much value. 5. JTBD (Moesta) lends itself to overthinking. Sometimes customers really just want a quarter inch drill. You can use JTBD to make it things way more complicated than it needs to be. 6. JTBD (the Moesta interpretation) lends itself to wasting time arguing about stupid shit. It almost never matters if something is an an anxiety or a habit or event 1 or event 2. 7. It's really easy to describe a JTBD so abstractly that a product person can't do anything with it. 8. There is a fine line between empathetic thinking and just making stuff up. Empathy helps you listen and learn while you are talking to someone. But keeping the empathy going after the conversation is how thoughtful people inject bad data into a project. About the Two JTBDs (and this is where I might butcher their work. 9. Moesta and Ulwick mean different things by JTBD. Ulwick's work is excellent at the functional side. Moesta's work is excellent at the social/emotional/long timeframe dimensions. 10. I have found it most helpful to use Tony's work to make the customer's JTBD very concrete: receive a payment, run a campaign to get leads. 11. I have found it most helpful to use Bob's work to make the customer's JTBD about a transformation or future state they desire. 12. Putting them together: a customer executes a core functional JTBD over and over with the idea that in the long they will achieve some sort of transformation. That's how the customer, task, product work together. As I shift my focus from consulting/researching back into making product -- I am using both to the fullest possible extent because I find that the customer data structured in JTBD thinking is impactful context for any agent helping me build.

Polish has its limits. Apple still rides the wave of goodwill they generated in the ~2006-2010 Jobs era when the products were thrilling. Now? They're fine. My limited experience with Omachy (2 weeks) has revealed it as easy and fast fun. Right amount and type of polish for my taste at least.

Decided to cut a good habit today The last few weeks I've read a couple of great books in the morning - ReWork by the guys at Basecamp and Show Your Work by Austin Kleon. They're both so great in helping me think about moving the business forward. But both so stimulating that by the time I was done reading I was so full of new ideas that it was a temptation to start pursuing them and add a bunch of stuff to the to do list. And my workload is already a step or two beyond workaholic. So cutting that morning reading and switching up to something far less applicable. Up next.

Hot take: just focus on working out when you’re working out.

This makes a lot of sense if you’re Iran. Where else would you go to get better advice on how to surrender?

Claude Code (allowing it to select its or model) for product and running business. Grok 4 for fast api calls Gemini 3 pro for thinking + writing api calls Have used OpenAI in a year. But I’d switch in a second if they come out with something better Zero reason to “switch”

Which will be amazing until you’re stuck in the infinite loop of “what’s a lead vs opportunity”. The more attractive outcome is that people who know crm will start vibe coding tools for 1000 company markets. These CRMs will be amazing to use and get results.

"SaaS is dead" is the conclusion he reached after spending hours vibecoding a copycat version of a product he could get for $9/month.

ChatGPT's model naming get clowned, but it's a normal rough edge for a product at this stage. I find it refreshing they're focused on product without caring much about branding. Enjoy the rough edges and the fact all the models are there to tinker with.

"I've got just enough low self esteem to get me where I want to go." -- Jimmy Iovine:

Don’t let the door hit you on your way out