May 01, 2026 · AI Code

Starting June 1st GitHub will switch from the current request based billing to usage based billing.

Meanwhile Anthropic has doubled it's estimate of the daily cost per developer for Claude tokens from $6 to $13.

I've been using Claude via Copilot and OpenCode for the last couple of months and rarely exceeded my monthly included requests with my biggest month costing me around $15.

Anthropic's latest estimate equates to $150-$250 per month per developer. It remains to be seen how this move will affect GiHub's pricing but it's safe to assume it's going to go up quite a bit from the current model.

April 28, 2026 · AI

ANANYA BHATTACHARYA writing for Rest of World:

The biometric ID project has been halted and investigated in multiple countries, but it recently partnered with Tinder, Zoom, and Docusign to verify users.

Even the fact that the company that launched the "World" project is named "Tools for Humanity" is gross, when what they are doing is building a database of peoples eyeballs. The "World" project was previously named "Worldcoin" and involves a cryptocurrency, with participants receive approximately $50 worth of the Worldcoin token for allowing their eyeballs to be scanned.

There's a good timeline of this attempt to build "A Biometric Database from the Bodies of the Poor" here.

There's also allegations of predatory tokenomics and insiders dumping their tokens on retail. This from Komando.com

The crypto token you get paid in? WLD launched near $1. It hit $12 at its peak. Today, it’s trading around 27 cents, and in March 2026, the World Foundation sold 239 million of its own tokens while the price crashed to a record low of 24 cents.

Companies partnering with World need to be boycotted [in my opinion] to stop this.

April 10, 2026 · AI

I used AI. It worked. I hated it.: Taggart Tech

There’s a fundamental problem with these tools beyond the capacity of any deployment strategy to solve: the tool requires expertise to validate, but its use diminishes expertise and stunts its growth. How does one become an expert? There are no shortcuts; there is only continuous hard work and dedication. I was once told of writing, great writers learn how to break the rules in new and ingenious ways by first learning the rules.

This is the same problem that I've been thinking about. Current software developers are able to verify and review the AI's work because they're experts. But what happens when people stop learning and acquiring the expertise?

March 27, 2026 · AI

Last year I started using Copilot. Copilot's inline completions were often wrong (they didn't have full app context) but autocompleting and refactoring is faster than writing everything yourself. Chat was useful for writing small methods or basically saving you the time it would take to go and scan Google or Stack Overflow. All-in-all good efficiency gains.

This year I started using opencode with Claude (Sonnet 4.5 and 4.6) and yeah, Claude can code. Claude is good enough to write features on his own.

I want user registration that verifies the users email, make the link valid for 7 days, add forgot password, and enforce password complexity.

Claude can do this in a couple of minutes.

My experience so far is that you want to keep features reasonably small, review, provide feedback, ask for changes. Treat Clause like a less experienced developer on your team. And don't get carried away and let Claude write too much without stopping to properly review and understand everything.

But Claude is getting good and development is changing.

March 18, 2026 · AI Apple

If they can make Apple Intelligence a first-class agentic AI by relying on Gemini, paying only $1 billion per year, it sure looks like genius. But given their track record with Apple Intelligence to date, that is an enormous “if”.

Only time will tell if this is really genius strategy, a happy accident, or a big miss. Or maybe a happy accident will be indistinguishable from genius strategy and we'll never know.

What I do know is that if we can get intelligent Siri, with all the intelligence of Gemini, but with the privacy of Apple's cloud and on-device models, that'll be a big win for customers and I'm excited about that.

March 16, 2026 · AI

I don't trust the predictions that software developers will be obsolete. The culture of Silicon Valley encourages this kind of chest thumping. On the other hand, the predictions for PCs and the web, the big things of my career in tech, were similarly bombastic, but they were wrong. The web was huge, just not in the ways people thought it would be.

This matches my own thoughts as well. The people closest to the tech have historically been very bad at predicting its impact and adoption.

While software development has undoubtedly been changed forever I don't see it going away.