This setup has been working great for me. The chat and web browser models are both using the Ollama cloud. This is about as good as it gets, I think, without paying the Frontier model pricing. Chat: qwen3.5:cloud Web Browser Model: qwen3-vl:32b-cloud Embedding: sentence-transformers/all-MiniLM-L6-v2

Hey everyone, Big one today. Agent Zero is officially v1.1. Why now? Because the Plugin System changes the nature of the project. Agent Zero is no longer just a framework you run. It's a platform you extend. Plugins are live. Install community-built extensions directly from the UI. Or build and share your own. The Plugin Hub is open. We're shipping faster. We've restructured how we develop and release. Expect more frequent updates, starting now. Every plugin you build makes Agent Zero more capable for everyone. v1 is the starting line. Let's build the future of A0.

Hey peeps, can someone either advise or point me to a video that explains how to get my agent zero to remember convos from previous chats when starting a new chat? Thanks

I built an openbrain-mcp in rust.. inspired by the concepts in mr jones' vidya https://gitea.ingwaz.work/Public/openbrain-mcp This is a bit over engineered. I wanted something that was hybrid semantic search + keywords. I also wanted it to be hostable on a vps so that I can tap into the mcp from anywhere. You could use this a starting point for your own openbrain or just install this one. Its very snappy because of the design decisions I made. Enjoy https://www.youtube.com/watch?v=2JiMmye2ezg

I was skeptical about openclaw. It sounded like it was too powerful and dangerously so. So I ran it. I gave it a device and a name. I’m honestly impressed at how capable it is. I’m already spoken for. Agent Zero has been my one and only agent for over a year. I built my system to make work and fun stuff more to my taste. It’s still powerful but it’s not dangerously powerful. I’ve ran huge cloud agents and small agents locally and it’s just a fact that models matter. It’s said that today’s software would be useless without today’s models. I’d absolutely agree. But what about today’s hardware? Today’s hardware is absolutely playing a role in outcomes. Regardless of what outcomes AI come to they have to live somewhere. My agents live next to each other on my desk but in completely different environments. Agent Zero lives in Kali Linux and inside of a Windows docker container. Openclaw is autonomously residing inside a Mac Mini. I haven’t had a Mac in forever. It’s nothing like what I get from Windows. I appreciate all three OS’s for different reasons. They all have strengths and weaknesses. After a lot of researching and waiting for the other shoe to drop, and it has for many users, openclaw is operating and in my control. I recommend using both. Agent Zero is enterprise certified. It’s secure and fully functional at the same time. Openclaw is powerful when given its own machine. Giving it a vps is a waste of potential. Can Agent-Zero do things that Openclaw cannot? Absolutely. Can Openclaw do things that Agent-Zero cannot? Absolutely. Most abilities depend on the environment that the agent is in and how the user set it up.