New on the Anthropic Engineering Blog: Demystifying evals for AI agents. The capabilities that make agents useful also make them more difficult to evaluate. Here are evaluation strategies that have worked across real-world deployments. anthropic.com/engineering/de…

One other big advantage of LLM assisted coding is lowering the barrier to new technologies. Working on a macOS app and really wanted to try tauri but this would be a ton of overhead for me. But with LLM can pretty quickly start learning and implementing the basics

Yes… this would be nice… wouldn’t it…

this is the dream.

one pattern I like with running 2 coding agents is one focused on logic and expansion and generally "slower". While the other focuses on beautification of code, UI, docs etc. Generally "faster" loop so you always have something to do and focus on.

Building the tool that builds the tool (and complies with claude code ToS)

If I had to bet on one team for open source general purpose OCR its LightOn. HUGE release.

that's crazy... miccheck-chess.vercel.app

These are the hidden gem of the ai sdk. Buried in their docs but takes so little to make an incredible looking UI with them. Highly recommend

5 places and 58 cycles to go....

let's. go.

Ryan D’Onofrio Raindrop LFG!! so excited

excited to have ryan on the team :) (p.s. we're still hiring)

Yeah. Raindrop.ai

I am two days in and there is so much I want to learn and try and do and see and build.

There are so many problems that can be solved by simply giving an agent tools to DO work. Software you’d spend months perfecting collapsed down.