John J Masse

Adapted from Dan Shapiro, “The Five Levels of AI-Assisted Software Development,” 2026. Level names and descriptions rephrased for internal use in businesses. Level Name In one line L0 Manual People do the work; AI is at most a reference. L1 Assisted AI handles discrete sub-tasks; people drive and

The frontier model companies aren't your only option. The economics of self-hosted AI are already better than most people realize. Jensen Huang was at GTC 2026 and said something that got a lot of attention. He proposed that every engineer at Nvidia should receive a token budget

You're building software that would never have existed without you. Here's how to make it a better experience. If you've been following my recent posts about software zones and the difference between vibe coding and software factories, you might have walked away thinking I

Vibe coding is powerful. But power without context is how things break. Here's a framework for knowing when it's enough — and when it isn't. The level of governance your software needs depends on what happens when it breaks, who's affected, and how

You're not broken. The situation is just genuinely hard. Here's what you can actually do. I want to talk to the engineers and engineering managers who are carrying something around right now that they haven't quite figured out how to say out loud. Maybe

The gap between "I made a thing" and "I can sustain a thing" is where most of this conversation falls apart. Something is happening right now that I think deserves a careful, honest conversation. And I want to have it without being dismissive, because the thing

New research confirms what enterprise engineers already suspect — AI reliability is a governance problem, not a capability problem. A couple of weeks ago I wrote about the Autonomous Maintainability Index — this idea that your test suite and documentation might be the most important predictors of how much AI can realistically

I've been building software for over twenty years. In that time, I've watched frameworks come and go, survived rewrites that should have killed companies, and developed a pretty reliable instinct for when something is hype and when something is real. That instinct is confused right now.

Your test suite might be the most valuable asset you didn't know you had. I was sitting on my couch this morning watching a YouTube video about open source software potentially changing forever. The video was from ThePrimeagen — if you haven't watched him, he's

How heuristic-based prompting creates AI systems that adapt, learn, and handle real-world complexity TL;DR: I am curious to know if you converted your instructional-based prompt into a heuristic-based prompt, would you experience improved results, as I have? I have prepared this GPT to quickly convert