I’ve been running product & engineering for 20+ years and AI 🤖 is the biggest thing that’s happened since the internet. We’re at the cusp of transformative change across every industry/vertical & the potential for new innovation is truly staggering. For the first time in over TWO DECADES we have the chance to rethink 🧠 how product and engineering teams operate. We need to rewrite the rules and create a new playbook for the next generation of builders. It’s an incredible opportunity for businesses & technologist and will be filled with twists, turns and breakthroughs! 🚀 #ai

"Where are my childhood heroes?" we wonder. It's you. You're the hero.

AI as a tool is now approaching “super-helpful” status. It’s not just a clueless junior intern anymore. It can actually reason and assist you with real problems. But you still need to ask the right questions & set context well. And that’s a skill a lot of people don’t have. So how do you develop this skill? Well there is a very slow way, but its the best way. READ. That’s right. Read. Reading helps you write better questions & describe context more effectively. If you haven’t read books in a while start now. Reading is one of the essential ways to develop your skills to communicate with AI effectively in the future.

Listening to @templeofthedog tonight. Mostly forgotten probably, but for a short period of time was a great band.

Kevin’s mom died (from home alone). Catherine O’Hara was so naturally funny. But in home alone she was all of our mom trying to get back home to save us. What a great lady. RIP.

Sometimes you're up, sometimes you're down. The same goes for everyone. So don't count yourself out (or anyone else). You can always come back.

AI isn't going anywhere even if the bubble pops. People talked crap about the Internet too at first. And its bubble popped too. Then it proceeded to generate Trillions in value. Beware the naysayers.

America the beautiful. We have to return to that.

You know what's sad? What's happening to America. There's no unity anymore. Remember 4th of July in the 80s & 90s? It was the best - everyone together. Now everyone is so angry. The U.S. is one of the greatest nations on earth. Life isn't easy but this country still has a lot of positives. Hopefully we can keep making it a better place.

Startup rule #273: if you start getting good product-market fit double down on what you're doing and don't try a new half-baked strategy.

Did Google's Gemini 3 really leapfrog past ChatGPT? The cold, hard truth is... No. Not for most users. Not even close. It's clearly just an incremental improvement in terms of text. For example, it still gets a lot of things wrong. Ask it to build a slide deck for next year and it'll plaster "2024" all over the place. But while LLMs seem to be offering diminishing returns on the text front, they are getting more amazing at media: images, video, voice, etc. For example, Googles Nano Banana AI produces much more accurate, hyper-realistic images & video than past versions. But back to text, Gemini definitely is not beating Claude on coding tasks anytime soon, for example. The bottom line is that most COMPANIES don't need images or video, they need TEXT, so from that lens Googles release of Gemini 3 was a dud. And If text is plateauing that might mean throwing billions at bigger language models may not be the right answer. Other AI domains might show higher ROI.

Here’s why AI won’t scale… With a traditional, non-AI app a dev team can easily support 10s or even 100s of thousands of users with 1 web server costing $400/month at AWS or Azure. With an intensive AI app a dev team can only serve a max of like 15 users with 1 web server — and that’s it. No one has figured out how to scale AI yet & that’s why it’s so expensive. In the meantime, the AI companies are hyping up the tech, luring in large Enterprise customers, then hiding the insane costs until it’s too late. What’s needed for long-term cost sustainability is a technical breakthrough either with hardware (GPUs, etc) or software (LLMs, etc). And the AI companies ARE working on it. But remember, it took 11 years (2012 to 2023) to come up with ChatGPT. With that said, AI as a paradigm for doing work isn’t going anywhere. It WILL infiltrate every sector. But scalable COST EFFECTIVE AI is stil a ways off. The good news is that all the money that’s going into AI will speed things up so it won’t take another decade to make AI scalable. 5 years give or take should do it.

You’ll succeed if you deliver value to your customer consistently. But… Give them no value and you get nothing. Give them value they can get elsewhere and you get nothing. Give them value they don’t need at the time and you get nothing. Give them value that costs them too much and you get nothing. Giving customers value is actually very difficult. People talk about it a lot but doing it is another matter. Consistent value delivery is achieved by very few.

Some uncomfortable realities of increasing your earning potential: Do risky, disruptive or new project ideas Try more experiments than others Set difficult targets & try to meet them Get into higher optic meetings Build stronger teams by exiting weak team members Get off failing projects faster Even when you fail position it as a learning opportunity

Ever have those days where you're just mentally exhausted but physically fine? What do you do to get out of it?

The history of technology isn’t a story of replacement, it’s a story of people finding new ways of creating leverage. 25 years ago they said Java would be the last language we’d ever need. 20 years ago they said Open Source would kill dev jobs. 15 years ago they said Cloud would replace on-prem (it hasn’t). 10 years ago they said Bitcoin would replace money. 05 years ago they said Low Code/No-Code would kill dev jobs (again) 03 years ago they said AI 🤖 would replace human creativity. They’ve been wrong every time — just not completely wrong. Each wave shakes the world, then becomes part of it.

Starting a company and running it means dealing with a lot of boring work. Most of building a business isn't glamorous. It's easier to be a 9 to 5 exec, have 1 great Board meeting a month, and go home to tell the spouse. Owning a business is totally different. Once in a while you have a glorious day of victories like getting a big deal or launching a product but those are rare. Most of the work is tedious and you have to stay on top of it to be consistent. And lot of it is also very quiet. No one else can really help you on founder-level work.

Figure out your priorities and then organize your week so most of your time is going into priority 1, 2 and 3. Contrary to popular wisdom I'm not sure most people can really "win" there week - too much variance. But winning your week is totally possible.

Most enterprise SaaS users above the age of 40 don't want to use a chat interface.

Long term thinking doesn’t always apply. If you’re going to exit in a year then think short term. But if you’re growing an org from $30M to $100 then figure out long-game.

Willpower is doing the right things in the difficult times. Builders know this. It’s incredibly hard to build anything. Even if you have money it is NOT easy. Willpower matters a lot.