OpenClaw took the internet by storm. Or at least the portion of it that is actively following AI technology and trends. I've had friends and coworkers that hadn't heard of OpenClaw (or Clawdbot or Moltbot) until I brought it up recently. Which is probably for the best. Because it's not a technology that's secure enough out of the box for wide use. Hell I'm even scared to install it on my own machine. I've only run it on a Digital Ocean VPS.

That said, its explosion in popularity with early adopters and an interesting set of use cases is definitely setting us on the path for all-encompassing personal agents.

That's why you're seeing Anthropic giving some of its most interesting features to Claude, or OpenAI acquihiring the developer behind OpenClaw. Even Meta got into the game by purchasing Moltbook, a social network for OpenClaw agents, for reasons to be seen. But probably something to do with the team’s work on identity for agents.

The real signal this week isn't who's building the next agent. It's who's building the next platform for agents. Nvidia compared OpenClaw to Linux and Kubernetes. WordPress opened the door for agents to publish content at scale. And if you zoom out, every major player is racing to be the infrastructure layer that agents run on top of. The land grab isn't for the best agent. It's for the operating system.

I won't buy a Mac Mini, I won't buy a Mac Mini, I won't buy a Mac Mini...

📰 Article
Claude Dispatch versus OpenClaw
The New Stack

Claude is trying to eat OpenClaw's lunch. Remote Control, Dispatch, messaging via Telegram, computer control. Anthropic has been shipping features that directly targets OpenClaw's core use cases. Dispatch lets you text Claude a task from your phone and have it spawn a desktop session to handle it. Remote Control connects your phone to a running Claude Code session. And Computer Use lets Claude fall back to controlling your mouse and keyboard when no dedicated integration exists.

🎥 Watch
How Anthropic Uses Claude in Product Management
Anthropic

This isn't a marketing demo. Lisa Crofoot, a PM at Anthropic, walks through how their product managers use Claude to query product data and build evals in minutes instead of pinging the data science team and waiting. No SQL required. It isn’t about replacing expertise, but extending what someone can do with their expertise. When you can ask your own data questions and spin up evals on the fly, you get to better decisions faster. Worth watching just to see how the people building Claude actually use it day to day.

🎧 Listen
Expertise Matters More Than Ever
The Aboard Podcast

Here's the pricing problem nobody's solved yet: you can now build a working prototype faster than you can write the proposal for it. The Aboard team dig into what that actually means for agencies and freelancers. If you're selling services and AI has compressed your delivery time, this one's worth the 30 minutes.

📰 Article
Nvidia's Version of OpenClaw Could Solve Its Biggest Problem: Security
TechCrunch

Nvidia announced NemoClaw at GTC. It's an enterprise-grade wrapper around OpenClaw with security and privacy controls baked in. Jensen Huang compared OpenClaw to Linux, Kubernetes, and HTML as a foundational platform shift and said every company needs an OpenClaw strategy. NemoClaw is worth watching because it aims to solve the "I can't actually deploy this at work" problem that's kept OpenClaw out of enterprises.

📰 Article
Vibe coding SwiftUI apps is a lot of fun
Simon Willison

Simon Willison talks about building macOS apps without knowing Swift or firing up Xcode. He admits he didn’t verify whether the numbers his apps report are accurate. That’s the gotchya with vibe-coding that doesn’t get talked about. Things may look polished and real, but if you don’t verify what’s under-the-hood, it is just a glorified proof-of-concept. If you're thinking about where AI-assisted development actually works (and where trust breaks down), this is required reading.

Thanks for reading,
Jason