Tim Hulse

#eFoil #Bermuda

Part of the thesis at @inanimate_tech is agents are beginning to act in the real world, and to do that well they need to be able to run code Like, our personal AIs can now drive my desktop apps. Why can't they drive the products in my office and home too? Why can't I vibe code a whole generative personal app that makes use of all the electronics around me? It's hard to see how we reach that future: Smart home protocols are too limited and too high latency. At the other end of the spectrum, imagine I could get my agent to re-write the firmware of all my devices for whatever purpose I have a dozen times a day -- is that going to be reliable? Or safe? We open sourced our answer to this today... it's called Resident. Resident is a code sandbox for ESP32 devices. It's aimed at device developers, and it lets them bring infinite reprogrammability to their devices for all their end users. (We love ESP32: it's a unique microcontroller family as it is used equally by makers, new hardware startups, and in mass production. Plus it has on-board wi-fi and a great ecosystem.) Sandboxes let you expose only the capabilities you choose -- and then the code can do anything. Once the developer adds the sandbox, it goes like this: - User says their intent - The agent reads the device capabilities doc: what IO does it have? What are its characteristics? - Then the agent writes code... - ...pushes the code to the device... (we provide a back-end server and websocket connectivity) - ...and the code runs in the on-device Resident sandbox. Boom, hot reload device functionality 💥 We've been using Resident for the last couple of months for prototyping new products and new use cases. It'll be at the heart of the experience for our future products. What we've learnt already is that infinite reprogrammability is powerful... but weird! It changes product archetypes fundamentally. We're going to have to figure out this new world together. Which is why we're sharing Resident for anyone to use. It comes with the same stack of agent skills that we use to quickly bring up new device prototypes and craft new apps for them. We're looking forward to sharing as we build, and would love for you to get involved too. 👉 Resident resident.inanimate.tech - Learn about integrating Resident (compatible with esp-idf and Arduino for prototyping) - Get the agent skills to create and push new sandbox apps - Try it instantly using the browser-based simulator 👉 More... We believe that on-device sandboxes are a foundational primitive for AI irl. I go deep on this topic on my blog interconnected.org/home/2026/05/2… Here's the GitHub — I'd love it if you gave us a star 🤩 github.com/inanimate-tech…

Wow, you can now major in “The Blind Leading the Blind” for just $28 grand at Chicago Booth School of Business. That 3-day immersion should really help with the imposter syndrome!

DeepSWE reflects what I’m hearing from engineers better than any other benchmark. They took the hard path to build a good one.

There has been amazing progress recently in vaccines that fight cancer. Just remarkable.

🆕 This black hole could have formed as early as a second after the Big Bang! Read more: esawebb.org/news/weic2609/ or 🧵👇

@blader I think that there’s something interesting there-that it’s the scaffolding that’s smelly - but we need the scaffolding.

one pessimistic view of ai progress is that our collective ability to smell ai generated text seems to be getting better faster than the models' ability to avoid generating those smells

My intuition about geometry was catastrophically wrong. I never knew that, in high dimensions, spheres effectively disappear. By 100 dimensions, an inscribed sphere occupies a smaller fraction of its cube than a proton occupies of the observable universe. By 500 dimensions, the volume is smaller than what standard floating-point arithmetic can even represent. The equations are real. Here’s the wild horror hidden in higher-dimensional geometry:

@MatthewTortora_ The Artemis Accords are an international organization designed to ameliorate this problem.

Yes, a perimeter, for Artemis Accords Safety Zones... No not "territory"... I hope we withdraw or revise the OST soon enough so we stop dancing on eggshells, crafting policy using corpo speak to avoid saying evil words like territory and sovereignty, and avoiding the fact that, of course, it will be like every other instance of exploration in human history, and no, pretending that we live in John Lennon's Imagine wont actually work as a platform for international space politics.

codex's 2x usage limit promo ends on may 31st that's in 4 days if your usage already feels like it evaporates too fast, get ready for june - it's going to be rough openai is willing to give eligible enterprise customers 2 months of free codex usage but pro users are about to lose the extra headroom that made codex feel like a workhorse so i'm hereby creating a petition dear @sama please make the 2x codex limits permanent for pro subscribers - you know it's the right thing to do sign my petition by liking this post and sending it to your mom

@kr0der I’ve had to stop using it. Can’t be stopping got five hours, faster to do it myself.

guys we might need a Codex rate limit reset, there's a really big bug, like that bug you know bug right there's that bug in the app like the bug, big bug

Starlink-equipped buoys provided real-time video streaming in the Indian Ocean during Starship's twelfth flight test

Inspiring!

90% of students don’t even know how to open or save a Word document when they come to my class, let alone format text. “Digital native” generation is tech illiterate & Grandma Luddite Heather must teach them the ancient ways

Could you DM me the analysis using this prompt below on Codex? It would help me diagnose and fix the issue. “Do a read-only investigation of `~/.codex/codex-tui.log`. Do not delete, truncate, rotate, or modify anything. Produce a report I can send to someone else with: 1. exact file size, modified time, and whether it is actively growing 2. the first and last timestamps visible in the log, if any 3. top repeated error/warning/message patterns 4. the last 200 relevant lines with secrets, tokens, API keys, user prompts, and personal paths redacted 5. any obvious runaway loop or repeated subsystem 6. the exact commands you ran Do not include the full log. Save the report as `codex-log-triage.txt` on the Desktop.” Then send me only that `codex-log-triage.txt`, not the full log.

C.S. Lewis, this is what life boils down to at the end of the day

I still can’t believe @karpathy released this 3.5-hour deep dive on how ChatGPT actually works... for FREE. 🤯 Easily the best video I've watched on the topic. Swap your next Netflix binge for this. Then read the ace article below from Codez breaking it all down 👀↓

@_sholtodouglas @DimitrisPapail Very respectful - love the 4 minute miles analogy.

Huge credit to the OAI team for solving the unit distance problem with 5.5 - it is now my go to example that models can in fact pull together disparate ideas into new discoveries. As with all 4 minute miles, we had to try and cross it too! Turns out mythos solves it with a cute, simple proof. This implies some serious overhang in discoveries!

Mostly true. What matters is securing the long-term future of consciousness, both on Earth and other heavenly bodies. We cannot just focus on Earth, because there are irreducible external (eg massive meteor) and internal (eg global nuclear war) cataclysmic risks. The Moon is faster to make self-growing, but is more susceptible to problems on Earth. Mars will take longer to make self-growing, because it is so hard to reach, but is more secure from Earth disasters for that same reason. Both the Moon and Mars should have self-growing civilizations. Making this happen is the prime directive of SpaceX.

@kr0der Enterprise has been brutal