Simon Willison

What we just talked about a moment ago where you have agents then owning the cognitive decisions because that's the only way you survive an exponential increase is you get agents in there that are exponentially in charge of some of those decisions too.

But it also reminds me of when we covered from Simon Willison, like a week or two ago, about, you know, 10,000 lines of code that used to be like a metric that you would gamify or whatever.

But now it's a metric that you understand, like what's the cognitive load of what a developer could even understand is going through an SDLC on any given day.

And you actually have to optimize down to prevent it from running away from what your humans can understand.

Developers wanna be everywhere.

I think about, um, Simon Willison, who says now that most of his coding is done on his phone, uh, and I look at that as the future.

And so I'm thinking a lot about wh-- A- as I think about our product roadmap, I'm building for developers who are just getting started with AI, where their experience is like tab autocomplete in VS Code, and I'm also building for running 12 parallel agents in the cloud because your machine can't handle it and you're s- sending all your prompts from your cell phone.

If my original question was, did it seem sane that y- you and this little project could compete with the giants of the industry? Uh, your answer is essentially, "I was so excited about the space that I didn't even care.

They're running on different Git worktrees, so they're not stepping over each other.

But that's like a great way to see how the different models perform because you're right, the results are drastically different.

You mentioned Simon Willison before.

I- I've become a pretty active reader of his blog.

OpenAI is both prover and sole judge, and the Clay Institute has not ruled on whether a version with an added forcing term counts as solving the original problem.

The only outside commentary on record from developer Simon Willison skips the math and reads the episode as opportunistic credit-taking.

Spending millions of dollars and ten thousand agents in under four days on a rumor is now a lab's normal response, turning who publishes first into a resourcing decision.

Separately, independent evaluators are finding Astra's headline claims don't hold up as broadly as OpenAI's own benchmarks suggested.

So who's work are you most excited about at the moment? I find

the most high signal-to-noise work source on the internet for keeping up with all the movement in AI to be Simon Willison's blog.

He is a longtime software engineer and he writes very short but high-value posts with all the main news coming out in the AI ecosystem.

When there's model releases or other news stories, he covers them in a very accessible way.

I think I actually probably mentioned it before on the show.

But when Simon Willison was talking about combining MicroPython with Wasm and running sandboxes in the browser.

I had

that on the show recently.

And there's a third component of it that I forget the exact definition of.

I think it was Simon Willison that coined it a few months back.

And that is a key thing is trying to separate functionally the concerns of how your agents interact with the world and So they can't both read all of your emails and send random strangers DMs on Slack or Blue Sky.

Yes, so there is potentially a hidden malicious prompt on public skills or plugins, especially with the advent of OpenClaw.

Overall, the community showed skepticism about the rollout process, but welcomed the core idea of native ad blocking on iOS with a focus on Mozilla's dependency on Google funding and the state of browser ad blocking ecosystem.

Title, the ChatGPT Codex app bundles a full copy of LibreOffice, source, Simon Willison's weblog.

The post examined how the OpenAI Codex desktop app, now called ChatGPT, includes about 1.7 gigabytes of data in its cache folder, including a full installation of LibreOffice, Python, Node.js, and other binaries.

The article noted that these dependencies are stored in the app's cache and are used to access and manipulate Office documents, despite concerns about the app's large size and dependency bloat.