Stanford dropped CS329Z today. Engineering AI Agents. compound systems, build RAG/tool use from scratch, then frameworks, then how you actually evaluate this stuff.
gonna follow along online. looks pretty neat.
cs329z.stanford.edu/
“I think the only programmers who have to worry about this change are those who refuse to embrace, or drag their feet on, the progress brought by the age of agents.”
— Over my dead pencil
via covereader.comworld.hey.com/dhh/over-my-de…
This.
“a Rust codebase is far more maintainable with agents than a Python one. Types are a form of verification. They provide back pressure.”
— software doesn't need to be readable anymore. it needs to be explainable.
via covereader.comghuntley.com/readable/
One of the many things I love about @badlogicgames' Pi is that it can just fix itself and its extensions.
This morning it hit some "subagents plugin can't access codemode" BS. tl;dr
I just said "fix it."
Two minutes later, unblocked.
Got tired of this email from GitHub, and of slow, expensive macOS CI runners.
Then I remembered I have a capable Mac Mini at home.
So I made a skill that sets up self-hosted runners on Linux or Mac. Give it to your agent and save some money: github.com/markphelps/skills
Switching my dev workflow back to a GUI with Orca: github.com/stablyai/orca
It does what I want out of the box, and the mobile app lets me manage agent runs on the go.
Much better than herdr + Moshi, with its buggy interface and constant upgrade prompts.
Dots = flop
Muse = toy, and I'm not giving Zuck more of my data on purpose
Grok Bot = overhyped
Building your own life OS and agent orchestration system on @Cloudflare = priceless
Reddit, LinkedIn, and (somewhat) X are all trying to keep agents away from content their users wrote.
That's not a moat. It's a slow exit.
Users will go where their agents are allowed in.
Eric Schmitt: "Did you or did you not fall in love with a Native American woman who could talk to willow trees and paint with all the colors of the wind?"
Jack Smith: "That was John Smith."
Cloudflare just released Clef and Clef-flash, two Jev-compatible decision models with image input.
Both are faster than Jev but cost more per token.
Disclaimer: I work on Workers AI but had no hand in this model.
blog.cloudflare.com/clef-dec…
The numbers:
Median latency: Clef 209 ms, Clef-flash 39 ms, Jev 524 ms
Input price per M tokens: Clef $0.24, Clef-flash $0.09, Jev $0.042
Benchmarks are Cloudflare's own