
To combat the confusion in this article I disambiguate between every term evalship.com/articles/eval-h…

Build. 👾 evalship.com 🤯 extraheadroom.com @extraheadroom 👨🍳 nommer.ai

To combat the confusion in this article I disambiguate between every term evalship.com/articles/eval-h…

Everyone is building an "eval harness" now imo 4 terms keep getting mixed up eval = a test eval harness = the code that runs it agent harness = the thing being tested harness engineering = improving that agent harness and none of them tell you when your evals are bad ✍️📝👇

Introducing @evalship 👾 🛸 👾 Tests, monitors & optimizes every AI call your product makes We'll send a notification whenever your AI calls need your attention Built for startups that need answers, not infinite dumps of data / traces 1 prompt to get set up 👇

Ask your coding agent to get started with: Set up evalship.com/setup.md

Hate the game, not the player
"I like money, bro. I agree. It's blood money." Jacob Coxon, the researcher who sensationally quit Anthropic, in a new interview at Stanford. Pressed by Rob Reich on whether keeping his OpenAI equity hurts his credibility, Coxon openly admits he still holds it, likes the money, and agrees it is "blood money." He argues that giving it away would not change anything ---- Full video on "Stanford HAI" YouTube channel, (link in comment)

Building for agents comes with new and unexpected challenges apparently I need to tell them explicitly NOT to copy my work, otherwise agents will just go do that? with @Padierfind


first time for everything honestly kinda surprising Claude has never before fucked up like this for me

Evalship launch video storyboard draft oneshot by Claude + OpenRouter API pretty great already

Looking for folks that got burned when switching models or changing prompts in their products would love to ask you some questions about how you deal with it

I walked at least 269,049 steps in October 🏃 ??? running without a device 🤺 ??? fencing without a device 🚶🚲 269,049 life movement
I walked 408,570 steps in October 2026 🏃 388,632 steps on a treadmill 🚶 19,938 outdoor

“I’ve been using Headroom for about a week now, and it has delivered exactly what I hoped for. For the first time in a while, I got through a full week without running over my token limit. I didn’t have to ration my…” magenta4000, 5 tacos on AppSumo 🌮 Get Headroom as a lifetime deal: appsumo.com/pl/OfXsbj6GmV/


Evals. Anyone working with evals in their AI powered product or considering working with evals? Would love to ask you some questions.

Hit $20.000 in sales on AppSumo 🚀 grateful for the opportunity to distribute like this very curious how this will impact long term findability of Extra Headroom - would love it if this also has long tail effects!
