Everyone is selling AI Fantasies. Instead of owning the AI, we just rent it from others because it is the best out there and we only use 10% of that best, but we want the best.
Their AI brings 95% efficiency to your R&D Department, but you do not have one. Still you want it.
One task, one run, so this isn't proof GLM plans at Fable's level across the board.
But it's a marker for how fast the open-weight side is closing the distance. And open weights come with an availability story closed models can't currently match.
Full comparison: https://t.co/ctQ4Ik11f3
Creator of Sqlite on pull requests: "You say, oh, it's free. No. It's not free. What you're doing is asking me ... to maintain it for you, to to document it for you, to test it for you, to maintain it for you for the next 25 years. That's not free." Yep.
Wise words from a wiser man than me. I've told people for the past decade and I have recent posts on here saying the same: the merge button is the easy part. Its the decade+ (Richard says 25 years) that follows where you've accepted the transfer of maintenance thats hard.
@FredKSchott I've been waiting for a stable version. I have already started experimenting it to build PowerPoint Add-In. Works great and is straightforward to deploy. thanks @FredKSchott
The problem with the "if it works who cares what the code looks like" mindset for agentic work is that it assumes the agent has a perfect understanding of "works." Realistically, things are underspecified, agents make bad assumptions, etc.
To be fair, agents are pretty good at unit test coverage. They're pretty bad at designing human experiences (API, CLI flags, etc.), especially cohesive ones for future roadmap plans they may not have visibility into (unless your backlog is perfect and vision fully laid out, which I doubt). They're bad at knowing where performance matters and what type (CPU vs memory tradeoffs). They're bad at where compatibility matters and where it doesn't (and tend to err on the side of preserving it without further guidance). Etc.
Unless you have this ALL specified, you can't possibly claim "it works" without taking a look and thinking about it.
I made a /radio command for pi (by @badlogicgames & @mitsuhiko) that streams radio garden stations while you work.
Random station from anywhere, search by name, or browse by city.
https://t.co/5iIM6jgaIH
Built a simple tool on top of @badlogicgames's pi and macOS Automator. Select any text, hit ⌃⇧Q, and pi opens in terminal to explain it. Then keep asking follow-up questions.
13 lines of bash. Zero config. Works in (almost) every app.
https://t.co/4G4RoVPldv
With the announced Rubin architecture, a box the size of a household fridge would have a peak power draw equivalent to that of around 65 households (1 MW per rack density). This rate of increase in power density has no historical precedent in electrical engineering.
@lucasmeijer I am using pi in cmux which has its built-in browser. works well for me mostly. on certain use cases I still use external browser through playwright.
Claude for PowerPoint does not work with Bedrock Gateway in EU region models. Don't know why. It just won't work. So I built my own AI Slide Generator. It is a beast. It is crazy how good it is. I could've just built a skill, but I built the whole app. Every day it gets better.