Spending Time After Hours Talking About AI and Development
New models launch every week and every lab claims the top spot. Here is how engineers can use Artificial Analysis, and its Intelligence vs. Cost per Task chart, to pick the right LLM for production.
A Postman desktop update reportedly added plugins to users’ Claude Code and Codex setups, along with instructions for the AI. Nobody has shown data leaving the machine, but it’s a good reason to spend two minutes checking what your agent is running.
We pointed six AI security reviewers at the same deliberately broken app. All six caught every dangerous bug, and then they disagreed about almost everything else. What that means for checking your own projects.
Yesterday your AI could read your Google Drive. Today it says “needs authentication”. Here’s where the reconnect button hides in Claude Code and Codex, in the terminal and in the desktop apps, and what to do when there isn’t one.
If you’ve ever plugged an AI model into an app, you’ve probably hit this problem. You ask for JSON, and most of the time you get it. Every so often, though, the model adds “Sure! Here’s your data:” in front…
I believe the most underrated part of any design, especially now with AI showing up more and more in content creation, is design standards. Letting AI be a tool or an assistant for creating code or media without a style…
Why I made this Before I got into coding, I cooked for 12 years. In a kitchen you’ve always got a bunch of tickets going at once, and every dish gets its own spot on the line so nothing gets…
Your AI coding assistant forgets everything between sessions unless you give it a memory. We look at how memory works in Claude Code, Codex, Gemini and Copilot, why a shorter note usually beats a longer one, and a seven-line file you can copy tonight.
If you keep pasting the same background into Claude, a Project can hold it for you, so every new chat already knows the job. Here’s what Projects are, what they’re good for, what they cost, and one example you can copy tonight.
Most AI writes. Jev, a new model from TypeSafe AI, only decides, and it does it in about a tenth of a second for a fraction of a cent. Here’s what that’s good for, what independent tests found, and why “can’t hallucinate” doesn’t mean “can’t be wrong”.