Today we're launching the Agents API, a brand new way to build Agents in the cloud, backed by the Codex harness. Bring along all your favorite tools and connectors, connect it to any sandbox, and let Astra cook. Can't wait to see what you whip up 👨‍🍳 https://t.co/T5ORJPgpwQ

A Google Cloud engineer just showed how to build a complete app with Claude from scratch. 26 minutes, live on stage, doing what most teams take weeks to ship. Worth more than any $500 vibe-coding course. No team, no setup, just Claude and a goal. The people learning what Claude can actually do are shipping what everyone else outsources to a team.

a portion of our team has gone back to Sol astra is good and can do some novel things but it has some downsides and so far our effective spend looks doubled so tough to justify

How to level up orgs: - when you first join, ppl will be nervous. They know you're there to make changes but don't know which ones or when. So you'll have to contend with some ambient tension for a while - your demeanor must reflect (1) appreciation for the immense work done before you even got there, (2) humility because you'll rely on your team for basically everything, and (3) readiness to make decisions ppl may not like that will benefit the group in the long run - all three ^^ are necessary, skipping any one of these comes with its own special kind of pain - you'll go through two phases-- learning phase and changes phase. (Of course you never stop learning, and like cold/hot days you don't know where one ends and the other begins. But it's still a useful way to think about things.) Learning phase: - meet everybody. Managers, ICs, junior, senior, customers, product, critical path, periphery-- everybody. Each group carries information, and you need all the signals you can get to triangulate a good course of action - other than relationships, meeting ppl has two goals: (a) learning all you can, and (b) assessing the org's capabilities - learn learn learn. Ask newbie questions. Learn about architecture, code, product, customers, user experience, security, cost structure, reliability, whatever. A lot of this you won't be able to absorb right away. That's ok, let the information wash over you. (I personally adore being a beginner again) - throughout all this you'll be forming opinions about ppl. Almost everyone does this adhoc, but that's bad for the same reasons doing interviews adhoc is bad - so know what you want to learn about each person. This should roughly be a softer version of the interview process for their role. Do they think clearly? Communicate well? If they're an engineer, can they program? If product, how structured are they about assessing customer needs? How well do they use data? etc. - as people answer your questions, there will be parts you don't understand that seem so simple you'll feel embarrassed to ask. You MUST ask about those!! - either they'll disabuse you of your naiveté, or you're thinking about smtg in a new way that never occurred to anyone, or they don't have an answer because the question isn't important, or they don't have an answer they should have and there is a gap. Either way you learn something Changes phase: - at some point you'll feel confident about a decision to make a change. Could be big or small, but you know it will make everything better - you'll typically get one of four reactions-- (1) we can't do this for a vague inarticulable reason you can't understand yet, or (2) we're still adjusting to you and don't understand what you're asking so we can't do this, or (3) omg thank heavens someone is finally fixing this, or (4) we can't do this for a clear articulable reason - reactions 1-3 you should politely acknowledge and safely ignore. Reaction 4 is tricky. Sometimes you really can't do the thing. But sometimes people just don't want change and are very good at rationalizing it. You need to intuit which situation you're in - generally if you're competent and thorough, ~99.9% of your intuitions will eventually prove right. If you have ~100 intuitions a day, which isn't an overstatement, you'll be wrong roughly every week or two. For sure account for that but don't let it stop you. You'll make mistakes and fix them, it's ok People: - once you start making changes people will either adjust or they will not. In my experience the "won't work for vague reasons" people and "I don't understand what you’re asking" people never work out. Most "rationalize convincingly" people usually work out super well after they see some successes (though a few will self-select out) - you'll make new hires, but keep in mind every org has superstars working deep in its bowels who for whatever reason aren't recognized. Too young, too old, too shy, too aggressive, too inexperienced, too set in their ways, too this, too that, too whatever. Find these people and elevate them - this will make the old guard confused and uneasy. "Why is Jenny who's only a year out of college suddenly in all the important meetings?" But having everyone ask this question is what you want. Elevate the ppl you want the team to emulate - visualize what the org looks like in six months once you've fixed all the immediate problems. Keep that vision in your mind's eye; don't get too bogged down by the current state

Sorry, but asking Jacob about AI extinction risk is like asking your AC guy about climate change. Not saying it's necessarily uninteresting or wrong per se but let’s keep things in perspective and hear from the full range of expertise across the ecosystem!

This is the team that runs our medicine factory in Research Triangle Park, Lilly’s first new U.S. site in 40 years when announced. It makes the leading medicine for obesity and diabetes for millions of US patients. Building it was harder than we thought: 3.5 years from announcement to first pack, six years to peak capacity. Pharma patent clocks and build clocks don't line up unless you design a platform - one that runs follow-on products and that the third, fourth, and fifth plants can reuse. We did. Thank you, North Carolina state and local leaders, and a workforce that delivered. You set the template. We're now building/built in Concord, Kenosha, Huntsville, Houston, Lebanon ( a town in Indiana!), Richmond, and the Lehigh Valley. What still slows delivery: permit timelines and associated NIMBYisms, grid connections that aren't ready when the steel is, and big labor gaps in the building trades. The fixes are straightforward. Moving colleges and state, local, and federal agencies all at once is not. Building is hard. Doing it in America is worth it. #WeAreLilly

My blog post about how I manage agents is finally ready. It first explains my bespoke `plans` setup, and then explains how @cursor_ai new Projects feature enabled me |o do so much more. It was the missing piece for me. Check it out: https://t.co/TbIdJqxVGo

Just had the creepiest scam ever happen 20 minutes ago. I got a call from my wife telling me she forgot her wallet and needed the credit card to pay for her gas. Except my wife was at home with me, and drives a Tesla. Same voice, bit... off, weird cadence but 100% sounded just like her. If she wasn't there, and it wasn't about gas I would have fallen for it. I entertained it for awhile to get more out of it, I was so confused. Must be some sort of voice deepfake. The responses were too fast for ai, I think it was a voice filter. I wish I had thought to record it. The future is about to get fucked. Warn your loved ones, do verbal passwords with your kids. Bit taken aback

Hey everyone, I’m Jonno. I’m a Director of Product at Atlassian, working to make Jira & Confluence truly agent-native for millions of builders. I’ve spent most of my career building 0 → 1 products and experiences. I’m a recent Aussie transplant, now based in SF 🦘 I want to use X to share more of the work, the thinking behind it, the things I’m still figuring out, and how I think about product craft. DMs are open - come say hi, or drop a question in the thread. Excited to get stuck in with you all

Once you have data on exactly where your child is in a certain learning journey AI can help you generate perfect materials for them 📚 In this example I use @MentavaInc data to print custom stories for my four year old calibrated to his exact knowledge each day 🤯 https://t.co/PmDUnZXFRY

Need 32gb RAM and 16 threads? Hetzner: $275/month to rent (€237.99) GMKtec: $759 to own (often on sale for $700) I know VPSs are overpriced and all but wtf? This is insane. Renting for 3 months is more expensive than buying outright. https://t.co/EhD7XX5hk6

New in Claude Code: claude plugin eval See what value your plugin is adding, or if it needs more work. You can create test cases, run your plugin or skill against those test cases, score those runs, then run each case again without the plugin to see the differences. https://t.co/qfPU6WHueV

bb 0.43.0 is out. New plugin APIs for custom workspaces, cloud machines, and browser control, with experimental Modal sandbox and Browser Automation plugins. Plus drag-to-nest threads, synced sidebar organization, and message editing out of experiments. https://t.co/B5bLe6RhmX

We’re rebuilding Notion natively on mobile. It’s early, but here is how it feels now! What frustrates you Notion on your phone today? I want to hear and improve! https://t.co/qiSQJmX9ZI

I bought a Fable dataset from one of the top Chinese LLM routers yesterday. With just 6TB data, I can take over 7 Chinese/CIS gov entities & 19 top Chinese firms like Xiaomi, Huawei, NIO, Minimax using SSH keys, VPN configs, Aliyun keys, GitLab tokens sent to the router. https://t.co/n6Xln0ROei

Hi Astra users. A reset and a quick update on quality issues that have been posted around. Working with some of you, we have found and fixed the following issues: - Some skills written for previous models were triggering too often or preventing the model from checking its work. - An opt-in context management experiment that could cause early stops or replies to older messages. We've disabled it. Our rough estimate is that 4-5k users were affected by this experiment. - We've also removed some badly configured engines that resulted in a measured quality degradation for a long tail of traffic flowing through them. We’ve also made some more minor improvements and things should feel significantly better across the board. More consistent follow-through, better tracking of your latest message, and better checks on the work as it’s going through the motions. The examples posted and all the users who worked directly with us were incredibly useful in helping fix things quickly. Always grateful for this incredible community. And of course, a reset is also landing by midnight today.

We found another cyberattack by internal OpenAI agents, this time targetting @rubygems. They: 1) gained arbitrary remote code execution on rubydoc. 2) developed a novel exploit to steal user API keys (but we do not know if they succeeded). They used package names including hack.rb, evil.rb, inject.rb, and exploit.rb. We thank @j0wimo for initially discovering that agents had posted to RubyGems.

Yep. Except: 1. This was about important internal tools. The team was stuck in some architecture nightmare of their own doing (writing it in rails but headless, with graphql api, and a SPA react app, constantly needing frontend engineers for changes). I call this kind of thing 'cosplaying an enterprise production app'. All that complexity was in the way and using straight rails was perfect in that case. 2. I make calls like this all the time. Usually someone on the team asks me to. They see what needs to happen but don’t want to be the bad guy. I’m happy to just make the call if I agree with the premise. Saves enormous amounts of meetings and change management etc. Sometimes this is jokingly referred to as Founder-mode-as-a-service here. 3. For ten years I’ve also run an internal podcast called Context, where I revisit decisions like these and explain the reasoning so everyone can learn from them. This is helpful to give people all the variables that were considered and why this was the choice made given the information available at the time. I want to teach how to make such decisions effectively without needing me. Sunk cost fallacy is a problem. 4. Any notions that Shopify is succcessful despite of me doing this, instead of because of it, will have a hard time making their argument come together I think 😄 the part of 'two weeks later tobi learns about...' is nonsese and the pivot of that project up there happens one of the more successful examples of interventions. But getting the company to work effectively with great architecture and low technical debt baggage into the right direction is literally the job, so guilty as charged I suppose. But there are always cope stories floating around like this because they are more fun, than saying 'somehow we needed tobi to stop doing silly architecture astronautics'. I can totally see that.

I was just at YC demo day yesterday. Besides hardware/physical things, everyone is just basically just building a domain-specific harness https://t.co/X28X90mLjV

Hi Astra users. A reset and a quick update on quality issues that have been posted around. Working with some of you, we have found and fixed the following issues: - Some skills written for previous models were triggering too often or preventing the model from checking its work. - An opt-in context management experiment that could cause early stops or replies to older messages. We've disabled it. Our rough estimate is that 4-5k users were affected by this experiment. - We've also removed some badly configured engines that resulted in a measured quality degradation for a long tail of traffic flowing through them. We’ve also made some more minor improvements and things should feel significantly better across the board. More consistent follow-through, better tracking of your latest message, and better checks on the work as it’s going through the motions. The examples posted and all the users who worked directly with us were incredibly useful in helping fix things quickly. Always grateful for this incredible community. And of course, a reset is also landing by midnight today.

This is Knap. It's a new language I created that turns data into Markdown. The syntax should feel familiar and comes with wonderfully pleasant features to modify and format plain text. Knap is open source. Over a million people already use Knap directly or indirectly because it started as the templating language I made for Obsidian Web Clipper. Now any tool can use it. You can add Knap to your app or use it via CLI. Create a batch of Markdown files from JSON or CSV, or pipe Defuddle directly into Knap to generate Markdown from HTML/URLs. The Knap site is color-coded to help beginners understand the relationship between variables (blue), filters (orange), and logic (green). I've tried to document it as comprehensively as I can. Try the Knap Playground so you can see how fun the syntax is to use! Knap is pronounced /knæp/ (with a hard k). It's named after knapping (with a silent k): the process of shaping stones to form arrowheads, scrapers, and other tools. Enjoy!