Any sufficiently advanced physics experiment is indistinguishable from magic:

CLAUDE FOUND THE SEEDS anthropic.com/research/ecdsa…

Sometimes life >> art

Martin is absolutely amazing! Any company would be lucky to have him. But whoever hires him, can you please let him work on @AIFoundryorg open source projects? I’ve got some you maybe interested in @marty188586 :-)

We're hosting our first technical meetup in SF on October 26. Drinks and pizza will be served, thrilling talks will be given. Join us: luma.com/0qr6bxcw

Before going to bed in 1826: make sure the fire is topped off with wood so the house won’t be completely freezing in the morning. Before going to bed in 2026: make sure the agents are topped off with queued work so the machines don’t sit idle all night.

Me want carry rocks sounds exactly like some of the software engineers I know:

I really need @romanvolobuev review of Digger. He’s the only one I trust on something as important as that!

I feel like Rust peeps stumble on something ingenious: if you want to protect your community from AI coding -- just make your compilation (or in general tooling) times so unbearably slow that AI agents won't be able to make much progress iterating.

ok this one is a f*cking goldmine 9 jobs where Jev engineering replaces a pricey LLM call (TypeSafe's full list) Jev is TypeSafe AI's first System One model, built to decide 100x faster and cheaper than a frontier LLM each of these 9 is a job you're probably still paying Claude to do move them over and this list will change your life the AI bill shrinks, and in the right hands the work you skipped as too slow or too pricey is what gets you noticed:

Y’all fools using it with Claude? Kimi is WAY more effective with it:

Jev as a judge for LLM evals! JevEval uses Jev as the judge for custom LLM evals instead of relying on another generative LLM for every evaluation. The idea is pretty simple. You define the exact decisions you want the evaluator to make, Jev evaluates each one and returns probabilities, and DeepEval uses those results to calculate the final score. For example, with an AI agent you can evaluate the trace itself, checking whether the agent selected the right tools, followed the expected sequence of actions, and actually completed the task. Jev handles those semantic judgments, while the scoring logic stays outside the model. That is the main difference from a normal LLM-as-a-judge setup. With a generative judge, the model often has to interpret the rubric, reason about the response, and generate the evaluation itself. With JevEval, you define the decision structure upfront. You can also assign different weights to different checks or use `strict_mode` when every requirement needs to pass. So if one part of the agent workflow matters more than another, you can reflect that directly in the scoring logic instead of leaving the entire evaluation to the judge model. Key capabilities: - Evaluate agent outputs and traces with Jev - Define custom checks as bounded decisions - Use yes/no, score, and closed-choice questions - Weight different checks or require all of them to pass - Evaluate both single-turn outputs and multi-turn conversations This is where Jev makes a lot of sense for evals. If you already know what needs to be checked, you can define those decisions upfront, let Jev handle the semantic judgment, and keep the final scoring deterministic. I’ve shared the DeepEval post in the comments.

$52M seed round. 15M+ users. But this is what we're most proud of yet. Introducing Drama 3: the most emotive and controllable TTS model in the world And yes, it's open-weight. You can now control how even a single word should sound, change its tone, and shift your voice mid-sentence. Uncle Roger was impressed too. Try it here: s.fish.audio/rqh0l8 Comment “DRAMA” within the next 48 hours for 1 month free. If you're an enterprise your first month is on us: s.fish.audio/qz9r0t

Watching this space with the industrial amounts of popcorn:

The hardest problem in personal kernel and alternative OS development has just been majorly simplified. You want basic device drivers for your local dev machine? Just drop a picture of the lspci output to the screen.

some fun stats from SF Tech Week - the largest ever - with thousands of events and 50k+ attendees. We've more than doubled since we started hosting in SF Amazing to see FiDi emerge as a force! Used to be that tech was just in SOMA. Plus the Mission+Dogpatch as well

openai just released Decisions API, Jev-like "system 1" multi-modal model based on GPT6 Luna! Each egg gets cropped and sent to OpenAI (~270ms latency) and returns P(clean, dirty, cracked). One image crop consumes ~280 input tokens (~$0.03/1K imgs). @roboflow RF-DETR for detecting eggs

Very disappointed in @jonstewart for giving platform to a fringe, delusional and frankly pitiful kid Jacob Coxon and presenting it in a way that papered over all those traits just to get a story. He could’ve gotten @bcantrill onto the show and what a show it would’ve been!

Good thing I am on exe.dev so kimi is /model away

Enemy of what tho? Of whom? Also, thank you mr. President for putting SIlicon into AI!

So it begins..,