For months -- maybe longer -- I've been the person my business partners and collaborators come to when their AI tools aren't behaving. One partner was arguing with ChatGPT (or Claude, or Gemini -- the specific tool hardly matters) about it not doing what was being asked. As we dug in, I realized our experiences were wildly different despite using the exact same interfaces. They were taking all the LLM's suggestions -- we are highly social animals, and it would be rude not to go along! -- whereas I am ruthless about keeping irrelevant and potentially distracting things out of the conversation. That one distinction transformed their results. The secondhand telling even improved their other business partner's chats.

Another collaborator wanted to prototype software without me hand-holding every step. I gave them a starter project, they used "vibe coding" tools to lay out screens and mock functionality over a couple weeks. We then spent a day or two rebuilding on the right foundation, and they were using it with live users the next week.

The push that finally tipped me over the edge came from my friend Andre Chaperon. I made a throwaway comment in Andre's community about sharing my thinking around all this "some day." Andre gave me a loving challenge: what was keeping me from shipping in public? My immediate answer was "time." There's a kernel of truth there, but it isn't the whole story. I then realized it was a form of perfectionism -- I didn't want to share incomplete thinking while things were still coming together. Which is exactly what Andre's phrase "shipping in public" is designed to inoculate against.

I also told myself a story about not wanting to be known as "the AI guy." I dismissed that just as quickly: I'm already being sought out for my opinions on this. I might as well share them in a way that helps more people. That's part of my calling -- to help the helpers, to amplify expertise -- so here I am, leveraging the same LLM-based tools to scaffold the work and share my messy thinking, making it clearer in the process of sharing.

What You'll Find Here

This site is a laboratory, not a showroom. Nothing is final -- especially considering how fast this whole field is evolving and the way tools and thinking are changing every day. [AI: This framing may need tightening -- I'm trying to capture the "laboratory" and "nothing is final" ideas you voiced. Feel free to rephrase or expand.]

My perspective is pragmatic, from inside the lab. I treat generative AI as an amplifier of existing expertise -- like a talented high-schooler who can do remarkable work, but needs careful quality control on anything outside your domain. I use these tools as thinking partners, not employees or interns. I'm firmly in the human-in-the-loop camp.

[AI: The following two paragraphs are my synthesis of where this site is headed, based on your seed list. Please review, edit, or replace with your own framing.]

You'll find me writing about how to keep the hyper-human aspects central -- making the important decisions yourself rather than delegating them to a model. About why provider independence matters when the technology moves this fast. About the vocabulary and mental models that help you navigate from simple chat completions through agentic loops and memory systems. About agentic coding, LLM-paired programming, and the levels of abstraction that let you work with these tools rather than for them.

And you'll find the frameworks I can't help but reach for -- parallels with human organizational structures, with how we learn and develop expertise, with the curse of knowledge and how it shows up in every generative AI interaction.

Who This Is For

This is for you if:

• You have expertise worth amplifying and you're looking for ways to do that thoughtfully
• You want to engage with the models themselves, not just the applications built on top of them
• You're skeptical of both the hype cycle and the doomer narrative
• You want pragmatic perspectives from someone who's building with these tools daily
• You're willing to do the thinking work, not just the doing work

If that sounds like you, I invite you to explore. Start wherever your curiosity leads.

---

This site is built in public. It will be messy, iterative, and evolving -- because that's how thinking works.

[AI: Would you like a sign-up or "raise your hand" CTA here? Or a simple invitation to explore the articles/resources?]