Rahil

update on teaching claude how to consult: the mechanics work. where it gets interesting is how it fails. cited withdrawn regulations as current. mixed up geographies. invented rate cards. claimed experience that never happened. got it asking questions before building anything now. analyst to associate upgrade, roughly. still aware... github.com/anotb/manageme…

pentagon calls Dario a liar with a god complex, blacklists anthropic, Sam signs a defense deal same day. if you ever wanted to look up what these people have actually said about... all of this. search it, chat with it. frontierarchive.com

Spent three sessions with Claude trying to extract a single corporate PowerPoint template into code. It placed a black logo on a black background. Hallucinated shapes that didn't exist. Took 15 debug cycles to get slide number padding right. 8MB of conversation. 350 lines of working JavaScript. Looked at everything. Claude Code skills, Cowork, Copilot in PowerPoint, python-pptx, PptxGenJS, raw OOXML. Every method breaks on corporate templates in a different way. And Microsoft hasn't shipped a sane API for creating presentations in 30 years. The fix is boring. Extract your template into code once, commit it, let the LLM just research and write content against it. unsol.dev/guides/llm-sli…

@statesman texas has 550+ data center facilities, more than any other state. 28 GW of total capacity. a lot of those planned sites are in areas already flagged for water stress. hard to blame communities for pushing back..

This Texas town shot down a proposed $1.5B data center project statesman.com/business/techn…

@YahooFinance the pipeline data backs this up. about half the announced capacity is landing in a handful of corridors already dealing with stretched grids.

Big Tech's data center push has sent electricity bills higher. Lawmakers want to slow them down. finance.yahoo.com/news/big-techs…

@zerohedge there are 15 facilities over 500 MW operating in the US today. the pipeline has 125. at that scale you're basically a utility, might as well act like one

Big Tech Turns To Uranium As Data Center Power Demand Soars zerohedge.com/energy/big-tec…

@business @tomwblack @opinion trend tracks with what the pipeline data shows. when the grid can't keep up with 250 GW of announced demand you just build around it. dcmap.us has the facility-level breakdown

"Bring your own capacity" can speed the AI data center buildout as long as it doesn't raise the price of electricity for residential users, @tomwblack says (via @opinion) bloomberg.com/opinion/articl…

@MarioNawfal the nuclear move makes sense when you look at the pipeline. there are ~125 planned facilities over 500 MW in the US right now. total announced capacity is around 250 GW. the grid adds maybe 50-60 GW per year. they don't really have a choice. dcmap.us

🚨 Big Tech is buying uranium mines to power AI data centers Microsoft and other tech giants are in talks with NexGen Energy about financing uranium projects. They need nuclear power for the data center explosion. Half of U.S. electricity growth through 2030 will be AI data centers. That's billions in infrastructure running 24/7. Solar and wind can't handle it. Nuclear can. So tech companies are doing what they always do: Vertically integrating. Can't get enough clean power? Buy the uranium mines yourself. Meta already invests in its own energy sources. Google too. Now they're all eyeing nuclear. The AI revolution runs on electricity. Tech companies just realized they need to own the fuel. Source: @ZeroHedge

built a dataset of every US data center. ~4,500 facilities at dcmap.us. there are 15 over 500 MW operating today. the pipeline has 125. median planned facility is 225 MW vs 10 MW existing. total announced capacity equals 5+ years of national grid additions. close to $700B in hyperscaler capex this year. half of it landing in corridors already dealing with stretched grids and water stress. capacity, pipeline status, climate risk, water stress, and energy mix for each facility. how much of this actually gets built is anyone's guess.

also an @openclaw skill. so your agent installs it, reads the skill file, and suddenly knows how to pull financial data without burning a gazillion opus tokens on API docs.

built a cli for pulling stock quotes, SEC filings, crypto, economic data from public APIs. got tired of juggling a dozen services with different auth and response formats so now it's one interface. also works as an AI agent skill so they stop wasting tool calls on API docs. github.com/anotb/open-mar…

@jelanifuel you're missing the daemon. openclaw runs 24/7 with its own memory system, messaging integrations, cron engine. claude code ootb starts and stops per task. pretty different animal

Openclaw hype is kinda wild to me. It’s literally just a wrapper to Claude code with more risk. The fact that it can run autonomously and call apis is the same thing Claude code can do… Not sure what I’m missing

@JeffGrayCyber @JorgeCastilloPr yeah residential proxies exist but you still deal with rotation, session management, providers getting burned. or you can set up a local exit node via tailscale and route your VPS through your home IP. both work, just more moving parts..

@Anot @JorgeCastilloPr Can a VPS agent use a virtual ip to hide it's nature?

Why do people buy Mac minis for OpenClaw when a VPS is infinitely cheaper?

@ExquisiteHabits @adamc0dez just ask it. say "i want you to be able to scrape reddit" and it'll figure it out. or look up repos on github where someone already solved what you need and point your agent at it.

@Anot @adamc0dez How do you add skills?

installed openclaw played with it for an hour honestly, it seems to me its useless it cant access reddit, linkedin, at least for me what should i use it for?

@yoavgo people confuse building with shipping. ai makes it easy to keep tinkering. the hard part is still deciding what's good enough and who actually needs it. and even harder to keep it working for others vs just yourselves...

everyone is apparently super busy having fun vibe building cool and complex software projects with AI. but why is almost no one releasing anything??

@championswimmer @openclaw oauth2 was built for 'is this client malicious'. agents need 'is this client careful'. different threat model entirely. you want gmail read to work, gmail send to need confirmation, gmail delete to be impossible. current auth can't express that..

The one thing @openclaw painfully makes clear is that we need an evolution of "authorization" models for agents. Currently there are two schools of thoughts on how to run Claw 1. Make it own accounts for everything including GMail, Whatsapp and everything 2. Run in on a PC where you are already logged in to everything so that it can inherit your tokens Almost all knowledgeable people of this space like @simonw have clearly advocated for sandboxing and have admitted they would never run it with full access to their life. Even @steipete recommends giving it aleast a separate number. But if you also make it a separate Google account, you now hobble a lot of its ability to handle your calendar, see your favorite places on Maps or summarise your emails for you. What doesn't exist (well technically does via auth "scopes") is giving Openclaw read only access to your Gmail without the ability to write. I think we need to evolve classic Oauth2.0 into another iteration where we acknowledge the existence of agents working on behalf of us. Just a reminder, Oauth2.0 already acknowledges the existence of "clients". In fact when you use Oauth2.0 you basically do the equivalent of giving your mobile app or the webapp your are using access to your account. Within that framework, we might probably need finer grained difference between when an Oauth token is being used for a fully automated LLM-at-wheels flow and when it is being used to give access to an app or website that a human is using. And then some sane defaults of what for-human tokens have access to and what for-agent tokens do.

@themgmtconsult The real threat to SaaS isn't vibe coding. Deploying open source got mass-market easy. What took a devops team now takes a few minutes with an AI agent. So the competition isn't users building from scratch, it's the free community (albeit vibe-coded) versions getting good enough.

Top-tier reality check. Have a detailed look at these two charts. People don't even want to cook! People don't want to do jackshit! If someone is willing to pay a 30% markup PLUS a delivery fee just to avoid boiling pasta, the idea that they will spend their Saturday "vibe-coding" a custom CRM from scratch is soooo incredibly batsh*t crazy I don't know where to start. If you think that's the future we are heading into, I have a medium portion of cashew nuts fried rice to sell you.