AI agents are moving beyond chat windows, and Mercury Agent is taking that idea to a whole new level. For a long time, we've interacted with AI by opening an app, typing instructions, and waiting for responses. Mercury is exploring something much more interesting: AI agents that become part of how we work, communicate, and manage our everyday lives. Its latest update introduces an extensible plugin ecosystem, and the first two additions offer a glimpse of where this is heading. Optimus: Mercury can see your screen, navigate applications, move the mouse, type, and interact with your computer. You can speak to it while it works, hear what it's doing, and stop it whenever you want. WhatsApp: Your AI agent is now just a message or phone call away. Ask it to call you at a specific time, receive updates and files, or approve actions through voice, even when you're away from your desk. But there's another important part to this. Each plugin runs in its own isolated process, comes from a checksum-verified catalog, and declares the permissions it needs. Sensitive actions like sending, posting, paying, or deleting require your approval. That's what makes this direction particularly interesting. We're moving from AI tools that simply respond to prompts toward agents that can take action, stay connected, and work alongside us across everyday applications. Less time managing an AI interface. More time letting an agent handle the work while you remain in control. It's a shift from AI as something you occasionally use to something that can genuinely assist you throughout your day. And the best part? Mercury Agent is open source and available for Windows, macOS, and Linux. Definitely a project worth exploring. GitHub: github.com/cosmicstack-labs/…

Don’t Push to Production on Friday.