This is a good “raise competitors costs” strategy Any competitor not doing this will look “unfair” I wonder what modelling they did internally for decision support on this - both the right thing to do and a clever commercial play

NVIDIA is rolling out Codex. And you don't let your engineers use AI tooling? lol, lmao even

Arguing Opus 4.6 vs Codex 5.3 slightly misses the point. GLM 5 is not that far behind either. It's not about the individual model it's about the overall mindset and way of working with AI. Are you letting agents figure things out or are you manually reviewing diffs?

One of the best ways to think through what the AI labs doing more for customers with forward deployed engineers, and all of the associated risks and sales cycle challenges that entails, is just read Palantir's 10-K risk factors section.

"We need to use AI more." More than what? For which tasks? How often? If you cannot answer, neither can they. brennanmcdonald.com/subscribe

The people who adopted AI—were they recognised? Promoted? If early adopters are not rewarded, why would anyone follow? brennanmcdonald.com/subscribe

If you have an idea, just try and build it with AI. How far you get is up to you. Just keep asking questions and climb the AI skill ladder.

if people don't feel like they will benefit from AI, they just haven't done the work yet a lot more people will catch up to what's happening this year

A lot of companies will struggle with AI projects in 2026. This is because they are rushing to get the benefits, without first doing the ground work. Clean data, clear governance, and just enough safety guardrails to not ruin the entrepreneurial trial-and-error this moment

The AI vision is clear to you. You have been thinking about it for months. Your team heard it once in a meeting. That is the gap. brennanmcdonald.com/subscribe

Since spending almost all my time in the last year on AI related matters, what I have come to realise is that many people are desperately scratching for any reason for this to not be real, so they can go back to "alignment meetings", "reviewing the code", and "working hard".

savings from efficiency gains generally get redirected into stock buybacks and executive compensation not much of the productivity gains from the last few decades has made it to the median worker these folks are correct to be scared, they saw what happened with outsourcing and

The irony of people worried about the risk of AI not taking action, simultaneously increasing the risk they become a victim of someone else taking action in their industry or that they never learn how to use AI properly so get let go for "not leveraging AI fast enough".

I think people who write about AI governance should be required to share evidence of how much $ they have spent personally testing AI tools and what their monthly "I am serious about AI" run rate looks like. You read some of this stuff and it's clear they are just theory.

You don't need any of it. Degrees. Courses. Certifications. If you are still in this mindset, you haven't put in enough work yet. That entire mental model is broken and failing and you can only figure out where to go from here on your own.

i feel seen

they have already figured out that everything in "prestigious" fields is too. damn. slow. the opportunity cost is too high - even 2 years wasted on wall st could mean permanent underclass for these strivers.

A lot of the human-in-the-loop and "taste and curation" stuff is pure cope. Have you ever read into time use studies? The daily usage of various apps? The doomscrolling sessions of *most people* outside of this X bubble?

Your team is doing the math. Effort to change versus risk of not changing. Change the math. brennanmcdonald.com/subscribe