The tipping point

Back in 2019 I ran into a silly interaction while pushing code to Bitbucket. I wrote a commit message locally Adding @font directives and pushed it up.

To my surprise, it had tagged the user @font , despite that clearly not being my intention. This was working in a private repo, so I was confused and a bit worried that I had implicitly given them access to this commit.

As it turns out, it was a harmless UI bug (original bitbucket bug report), and life moved on.

However, users would still get tagged if they were mentioned in the commit message of a public repo. So I decided to grab the username of the most popular css directive that I could find that wasn't yet taken.

And so my @font-face account was created.

I still get regular emails from people's public repos to this day, where I get little glimpses into the kind of public projects that people are working on.

As it happens, this is now somewhat of an interesting dataset to see the impact of AI coding tools.

Here are the repos that mentioned me, up until the end of 2025.

This felt consistent with Bitbucket being a dying platform, and petering out over those years as Github truly rose to dominance.

While it is a low volume dataset, and subject to variance, you can see that past the end of 2021 things have gotten quieter. There was a period of ~18 months in 23-24 where I did not get mentioned in any new repos at all.

So what changed.

For anyone in the industry, Claude Code + Opus 4.5 in November 2025 felt like the tipping point for AI coding.

So now let's zoom out, and see if the data supports that.

I have been tagged on new repos more times this year than in the previous 6.5 years combined. And the data only runs through the end of September.

In March 2025, Anthropic CEO Dario Amodei made a prediction that 90% of code would be written by AI within 6 months. Prior to November 2025, I was tagged in 0.54 repos a month. In 2026, I have been tagged in 5.4 a month. It took 18 months instead of 6 months, but at least on this small strange metric: I think that we may have reached that point.

Like it or not, the age of AI development is here.