Thoughts

A web app is text, code, and a database talking to each other. This is a hands-on walkthrough: edit real HTML, real Ruby, and a real database table right on this page, and watch how a request actually gets built.

Two questions sort every AI initiative: did the spend exist before, and is AI the engineer or the employee?

Your website was built for humans and Googlebot. Agents are the new visitors. Here's how to make your site legible to the LLMs that are increasingly reading it on your customers' behalf.

Every company building on OpenAI or Anthropic is making an implicit bet: that these providers will remain stable, affordable, transparent, and available. The evidence says otherwise.

A dollar-for-dollar breakdown of AI memory pricing, what models you can actually run at each tier, and why the Mac Studio is the most underpriced hardware in AI.

The strategy brain and the hacker brain, working together with AI leverage. Here's how we actually do it.

A technical walkthrough of identifying high-ROI automation targets and building AI systems that actually work in production.

A framework for evaluating AI readiness during due diligence — and how to price the AI opportunity into your investment thesis.

A practical framework for evaluating where AI can drive the most value in your business — and where you're not ready yet.

A step-by-step playbook for PE operating teams to deploy AI across portfolio companies — from diagnostic to full implementation.

Forget vague promises about 'efficiency.' Here's how to calculate the real EBITDA impact of AI initiatives.

Every business in the US, Google review sentiment, demographic trends, automated investment memos — built by two people in two weeks.

Three concrete AI implementations you can ship this quarter that will show up in your next board deck.

The economics of AI-native consulting: how infinite leverage changes who can deliver enterprise-grade work.

The old consulting model is broken. Here's what AI strategy work looks like when you strip out the bloat.

The benchmarks, the economics, and the case studies all point to the same conclusion: open-source models can handle the vast majority of what PE portfolio companies need AI to do. Here's the data.

The technology behind ChatGPT and Claude, explained by building one up from the simplest possible version.

EV/Revenue multiples across 40 major software companies are diverging dramatically. The pattern reveals which moats AI will erode — and which ones it won't.

I used AI to analyze every chord and melody in Weezer's discography. Here's what it taught me about business intelligence.

A plain-English explanation for people who run companies, not computer science departments.

I exported 7 years of Apple Health data, dropped it into Claude, and got interactive dashboards that outperform what any fitness app gives me. Here's what that means for SaaS, data portability, and the future of software.

You use one every day but probably can't explain how it works. Here's why understanding the basics changes how you think about AI.

Every process in your company was designed around human limitations. Time to start over.

No, but it's dying and nobody knows how quickly.

Every major leap in computing came from hiding complexity behind a simpler interface — from punchcards to programming languages to GUIs to the cloud. AI is just the latest layer. The winners will be the ones who see what comes after it.

The $600 computer that's replacing $200K employees. What this means for headcount planning.