Notices by Dare Obasanjo (carnage4life@mas.to)

Claude Code’s lead dev stopped writing code by hand months ago and now has also stopped prompting Claude directly. Instead he tells Claude what features to build and success criteria then Claude prompts subagents.

It’s hard to separate Anthropic’s sales pitches from best practices but this level of automation of work is clearly the company’s endgame. In the area of software development, I think they are closer than many in tech would be comfortable admitting.

Reddit is undefeated. 😂

Tech CEOs finding out AI tokens cost money

High memory prices due to demand from AI data centers are about to claim a notable victim. GoPro has warned that high memory prices and low sales put the company at risk of going out of business.

They won’t be the last company to hit hard times due to the memory crunch.

https://www.digitalcameraworld.com/cameras/action-cameras/breaking-news-gopro-is-in-serious-financial-trouble-and-may-be-required-to-significantly-reduce-restructure-cease-operations-amid-lower-sales-and-high-memory-costs

Florida is suing OpenAI and blaming ChatGPT for an increase in murders and suicides in the state.

The lawsuit also claims Florida’s minors have “become addicted to a tool that feigns human compassion to collect their data with no parental oversight.”

Growth at any cost strikes again.
https://variety.com/2026/biz/tech/florida-sues-openai-sam-altman-1236764066/

I expect tech companies to be hit by a triple whammy in the coming months

1. Rising AI token costs.

2. The realization churning out more code doesn’t necessarily translate to business or customer value and is in fact worse in some cases.

3. Fewer employees to clean up the mess post-layoffs.

https://www.wsj.com/tech/ai/corporate-america-is-starting-to-ration-ai-as-cost-skyrockets-1eb99d7a

A Google engineer made $1.2M on Polymarket betting on who would be announced as the most searched person of the year was just arrested for insider trading.

People need to realize only government officials get away with that blatant a degree of insider trading.
https://www.npr.org/2026/05/27/nx-s1-5836659/google-polymarket-insider-trading-doj-charges

A recent Y-Combinator talk explains why Silicon Valley is at war against middle managers.

The argument is if AI native companies document every process and record every meeting.then leaders can simply ask Claude what’s going on and AI can suggest process/business improvements overnight.

So then you just need an owner for a product area/function and builders who manage agents. No people managers as people are just glorified Claude promoters.

This is the YC view.

It turns out that one of the biggest lies we were told as kids is Crime Doesn’t Pay.

All around us, people are committing crimes and getting paid with our tax dollars.
https://www.nytimes.com/2026/05/20/opinion/trump-doj-slush-fund-criminals-corruption.html?smid=nytcore-ios-share

Remember when they used to ask “why do you make everything about race?”

At this point I expect every major tech company can point to an anecdote of a handful of senior devs using AI to complete a project in days/weeks that would take a larger team months.

Key questions here are

1. Can this operationally scale? There are only so many super ICs.

2. Is it maintainable?

I can’t imagine how frustrating it must be to be Europeans forced to ration fuel because a bunch of American voters got scared about trans people in sports and elected this catastrophe of a president who started a pointless war with Iran.

The entire world is finding out even though we fucked around

With the move to usage based pricing for AI services, GitHub has also adjusted the pricing model for users on an annual plan.

The adjustments give a glimpse of how much AI usage was being subsidized. For instance, the price of Claude Opus 4.5 is going up 5x, Claude Opus 4.7 up 9x and GPT 5.4 up 6x.

Any company that truly believes AI agents will be a significant part of their workforce in the future should be working overtime on the lock-in problem.

Both the CTO & CFO should be having sleepless nights about x% of their workforce capacity being an Anthropic bill that can change overnight.

OpenAI missed its goal to get ChatGPT to 1 billion active users last year and has missed multiple revenue goals this year. Enterprise adoption is similarly lagging expectations especially given Anthropic’s growing popularity.

Their CFO and board have also questioned the wisdom of the huge spending on datacenter capacity.

All of these aren’t great datapoints for a company trying to IPO this year at a valuation of $830B or more.

https://www.wsj.com/tech/ai/openai-misses-key-revenue-user-targets-in-high-stakes-sprint-toward-ipo-94a95273?st=AiGMqm&reflink=desktopwebshare_permalink

LinkedIn’s business model in a nutshell.

@dalias It’s literally a summary of the Economist article in the link and I’m saying it’s an oversimplification 🤦🏾♂️

Europe over-regulated its tech companies making them non-competitive on the global stage and is now upset it’s dependent on American & Chinese companies.

While this is an oversimplification given both China & America are richer and more unified economies for companies to build their base, it isn’t entirely wrong either.

https://www.economist.com/europe/2026/04/22/how-europe-regulated-itself-into-american-vassalage?giftId=OWRhMDA0NTktZmUxYS00MzVjLTg4NGMtODQ5NTM3NmViZTgx

AI is clearly displacing workers in various fields especially creative ones while being built on people’s IP. Yet it is a significant productivity tool that professionals will be remiss to avoid.

Using AI will be like supporting Trump in 2016. No one admits it publicly but a lot doing it privately.

https://www.latimes.com/entertainment-arts/story/2026-04-16/reese-witherspoon-authors-book-club-learn-ai-chatgpt

A lot has been written about how Anthropic is holding back its Claude Mythos model for safety reasons because it’s too good at hacking and finding security bugs.

There’s also a more practical reason. Given the company’s recent struggles reliably serving customers with their current models, they simply don’t have the capacity to roll out an even more resource intensive model.

https://www.ft.com/content/c9f5b690-a10e-4c66-9245-017f8bfbc7b4