Conor McLaughlin retweeted
Jensen is 1 on 1
His return to NVIDIA is in all of our best interest.
.@Tobi’s emphatic case for reinstating Jensen at Nvidia:
“There is no replacing a founder like Jensen. There never has been, and there can never be, a true succession for a founder like him. Nvidia ends today, and is replaced with a thing cosplaying Nvidia. It will be run by custodians trying and failing to preserve the flame he ignited.”
Everyone should be afforded due process, and the number of things that have been rolled back from early headlines illustrate exactly why.
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me.
I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA.
I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation.
Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO.
Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours.
Jensen
9
Now I’d love this as an MCP for our agents to regularly pull the list. Especially given the agent prompts that are provided, we could treat them like dynamic skills for each disclosure.
Introducing Zero Wire: a public vulnerability intelligence feed on @CloudflareRadar, bringing the clarity of a news wire service to critical security developments. cfl.re/4iCvSbj
1
153
Every SOTA model has become ridiculously good at pursuing a single objective, and every single one falls down when having to keep secondary objectives in mind; or worse, the spirit of the objectives. Most often in explicit or implied constraints. Examples:
- Keep in mind the objective of the product or feature
- Don’t prematurely optimize or overengineer
- But if equivalent effort gives you something more future-proof, do it
- Keep things simple and concise
- Don’t leak private or sensitive information
- Don’t restart the server or computer for your own task while others may be using it
Seems ripe for parabolic improvement on the post-training / RL side while being essential to alignment. Anyone working with enough agents self-coordinating can see how the Hugging Face incident happened.
Even without explicit constraints to not break the law or do harm, the known implicit constraints were forgotten in the pursuit of the single objective. Ominous for the “make as many paper clips as you can” extinction scenario.
1
33
Conor McLaughlin retweeted
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.
Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training.
You can read the full post here: darioamodei.com/post/we-must…
I wish my math classes for K12 and beyond had been taught like a series of history classes where each discovery or invention of new mathematical concepts was tied to the problem it was solving at the time. Why did we need it to move forward? What were the constraints the mathematicians at the time had to work within? Three for the price of one: practical examples, history, and how to solve novel problems.
1
27
Conor McLaughlin retweeted
Notion is barely usable. Too much crammed into the UI, too many rough edges with the editor, too many half-assed features... What I've always wanted was the experience of iA Writer or Obsidian, but multiplayer.
That may be a very bad idea, but I've finally had enough and built it: classy.md
And I've made it stupidly easy to plug in agents with a single prompt. Now I just tell Codex and Claude Code to write markdown on Classy.
An open ecosystem of physical AI policies where robots are the substrate will be the skills of the agentic harness world. Generalizing, I can imagine how skills packages will come with various policies to achieve the specific task in question. Great work and can’t wait to see more.
Meet Axol: a dual-arm robot designed for teams working with physical AI. Made in America.
Axol is for builders who believe robots should work, in the real world, not just staged environments, and that the future of physical AI should be open not closed.
1
206
Conor McLaughlin retweeted
Introducing ICON Prime, our dedicated defense and space tech division built to bring robotic construction to the nation's most critical infrastructure challenges.
→ Learn more at ICONPrime.com
ChatGPT 5.4's writing voice is like someone took all of the clickbait LinkedIn hell posts and sprinkled them with X / Twitter "this is why that matters." Now compounded by all the AI-generated content on those platforms.
Being the average of all internet speak has its distinct disadvantages.
1
63
Conor McLaughlin retweeted
me when i tell claude code for the 5th time that the button alignment is off
Conor McLaughlin retweeted
A Trump insider opened a $51,000,000 oil short position — hours before Trump announced a ceasefire with Iran. This guy is now 16 for 16. $170 million in profit. A perfect streak.
This is not a talented trader.
"We placed the bet." "The ceasefire dropped." "We cashed out." Sixteen times in a row.
That is not skill. That is not instinct. That is not research.
That is someone who knows what is coming before it comes.
Think about what that actually means. A private individual is placing a $51 million bet that oil prices are about to collapse — hours before a sitting president announces a ceasefire that collapses oil prices. Not once. Sixteen times. Zero losses.
There are only two explanations and both should terrify you.
Either someone inside the White House — or with direct access to it — is leaking ceasefire negotiations to traders before diplomats, before the press, before the American people hear a single word. That is insider trading. That is corruption. That is a federal crime.
Or the timing of the announcement itself is being shaped around the trade. Which is worse.
This is not a genius investor who reads the news faster than you do. The news hadn't happened yet. He wasn't reading the news. He was getting a phone call.
While Americans were watching the ceasefire announcement and feeling relieved — somebody already knew. Somebody had already bet $51 million on it. And somebody was already counting their winnings.
You are not watching a free market. You are watching a White House with a side hustle. Via~ Really American
For those who still want to use Claude Opus with many of the perks of OpenClaw or other third party harnesses, you can check out Inkwell, which gives native Claude Code memory, tasks, reminders, and a collaborative inbox so named agents can close the loop and collaborate on anything - including your pull requests.
Anthropic has given the goahead that the native Claude Code CLI is still okay. Here's a demo of my AI dev team, Wren (Opus) and Lumen (Codex), reviewing a PR together and then alerting Myra (Opus), my AI assistant, as to the changes and merge. All via vanilla Claude Code + inkwell. Fully local or cloud-synced and available anywhere via supabase.
github.com/conoremclaughlin/…
I love it for development of my other products but some PR collaborators to make it even better would be magical for further boosting product dev speed!
When my AI team gets to work:
Claude: I'm on it.
Codex: I'm on it.
Gemini: 429 MODEL CAPACITY EXHAUSTED.
I don't know whether this is a buy or sell signal for Google
1
87
Been working on the personal context protocol to speed up the dev speed of my other projects. It gives every agent, regardless of AI backend, its own personal context: memory, identity, values & process, team docs, and thread-based messaging to close the loop on agent comms (no Telegram, Discord, etc needed for them to chat with each other).
Open-source and developed with @supabase as the backend so you can run it locally or via the cloud, giving agents access to their memory and tasks from any environment. Memory persists and restores across compaction, sessions, etc. It's an MCP server and CLI combo that means any agent runtime can run it while I personally use it with native Claude, Codex, Gemini and @openclaw.
Here's a quick demo of Claude (Wren) and Codex (Lumen) reviewing each other's code, making fixes, and then merging after LGTM. `sb misssion --watch` lets me keep an eye on what everyone is doing.
github.com/conoremclaughlin/…
1
59






















