ericosiu

@trq212 wow

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.

@BioavailableNd As an American, I've never understood why Americans wear their shoes in the home. It's gross.

Wearing outside shoes inside the house. According to Dr.Gerba, Microbiologist, there's a 96% chance there's fecal matter on your sole and +400k units of bacteria.

The death of marketers, is it over for paid media?, the bear and bull case for AI x.com/i/broadcasts/1…

@Jayyanginspires @themaxyang @singlegrain I can tell you both had good parents. :)

@themaxyang @ericosiu @singlegrain @ericosiu Can 100% vouch for Max! He’s someone I trust deeply when it comes to building better workflows. Watching him go from zero to coding with AI - and actually shipping things that work has been awesome. He's a fast-mover and I'm proud to call him my twin brother!

Hey @ericosiu , I'd love to work with you as an AI Automation Intern at @singlegrain. Here's my pitch 🧵:

@themaxyang Ah just saw this will send to the talent team!

If this resonates, I'd love to build with you. My DMs are open! Check out the Library and my "Beat Claude" pitch here: notion.so/AI-Automation-…

I replaced a $27,000/month role with 4 AI agents. The output didn't drop. It went up. We're now averaging 85,000 views per X post. The AI agents run 24/7. Total cost: $200/month. That's a 135x cost reduction. I'm not sharing this to be cruel. I'm sharing it because this is the new math every CEO is doing in their head right now — they're just not saying it publicly. The person wasn't bad. The role was automatable. There's a difference. And here's what most people miss: the $27K didn't go into my pocket. It went into higher-leverage hires — people with judgment who can direct the agents, not people doing work the agents already do better. This is the shift. The companies that figure out the right human-to-agent ratio first win. Everyone else overpays until they can't afford to anymore. For more on AI, business, and marketing, just comment "newsletter."

@themaxyang @singlegrain Hey Max - love this Head to singlegrain.com/apply to beat AI if you haven’t already ;)

AI can already replace 95% of coding tasks. It's only doing 33%. Read that again. Anthropic just published a study showing the gap between what AI could automate and what it's actually automating right now. The numbers are wild. Programming? 95% possible. Only 33% happening. Customer service? 82% possible. 28% happening. Healthcare? 40% possible. Only 5% happening. Construction? 15% possible. 2% happening. That gap is the biggest opportunity in business right now. Every percentage point between what AI can do and what it's actually doing is a company waiting to be built. A tool waiting to be created. A workflow waiting to be automated. Most founders are chasing the stuff AI already dominates — chatbots, content, coding tools. The real money is in the industries where AI has barely started: healthcare, legal, construction, logistics. That 5% in healthcare? That's going to 40% in the next 3 years. Whoever builds the bridge gets paid. Stop building in crowded spaces. Build where the gap is widest. For more on AI, business, and marketing, just comment "newsletter."

Your competitor runs 50 marketing experiments a year. You can now run 36,000. For free. Andrej Karpathy just open-sourced Auto Research. It runs AI experiments every 5 minutes on a single GPU while you sleep. We’re already testing this at Single Grain. Here’s why this changes everything: 1.The hit rate is real. 83 experiments overnight, 15 keepers. At a conservative 12.5% success rate across a full year, that’s 4,562 winning experiments. 2.A junior marketer costs $5,000/month. This engine costs $112/month. 3.The compounding is what kills. Cold email learnings feed ad creative. Ad creative feeds landing pages. Landing pages feed discovery call scripts. 17 channels all learning from each other. I dropped the GitHub link into Claude and asked how it applied to our goals. It scored 92/100 on our leverage scale. Most ideas I test score low. This one hit immediately. The tool isn’t the point. The autonomous knowledge layer that gets smarter every night is. 50 experiments vs. 10,000. That’s 200x. No human team can outrun that math. What would you point this at first?​​​​​​​​​​​​​​​​ youtu.be/zgHoahKoqBg

What time were you sleeping before Claude Code/OpenClaw took off? After?

No, you can’t hand off AI transformation. It starts from the top. The most cracked out founders I know are sleeping far less (9-figure+). They’re deep in Claude Code and the Claws. All the hackathons, office hours, AI specialists, etc. won’t help if the CEO isn’t all in.

@girdley It’s already in there with a ton of safety checks

Temp check: Would you plug an OpenClaw agent into your business TODAY? Why / why not?

@seni0re @nvidia Yes it will

@ericosiu @nvidia This is gold! Do you think it will stay Open Source? What about the relationship OpenClaw+OpenAI?

Multiple 9-figure+ companies have asked for Claw implementation with caveats around security and HIPAA compliance. And rightfully so. @nvidia’s NemoClaw makes it all possible with enterprise security in mind. Such a smart partnership with OpenClaw. More below:

@MingtaKaivo wow

@ericosiu before: asleep by 11. after: still awake at 2am watching AudioWave's agent finish a feature I started at 9. the dangerous part is it doesn't feel like work until you check the calendar.