Hey all, would love to know what multi-agent systems you have setup and how they are doing? Loving Paperclip thanks @Nate Herk for the awesome video!

I haven't been super active here as of late, as I've been heads down building and working with clients. I built something last week inside Falcon Builder that I think the people in this group will appreciate. Falcon is my no-code AI agent platform — think visual workflow builder but purpose-built for AI agents, not just integrations. Last week I shipped AI Wingman — an AI assistant embedded directly in the workflow editor. Here's what it does: you describe what you want in plain English - it reads your entire workflow — every node, every connection — and proposes changes as reviewable cards. You approve or dismiss each one. The changes are surgical, not a full regeneration. The frame I keep using internally: it's the Cursor for AI agents. You describe the outcome, AI does the build, you stay in control. For those of you who build workflows for clients — the use case here is your client being able to say "I want to add an email notification whenever a new lead comes in" and having it done in 30 seconds without touching you. Built this because I was living the pain of it. Curious if anyone else has cracked this problem differently.

Anthropic's AI agent takes its first steps into autonomous computer use. Anthropic has launched computer control features for Claude on macOS, allowing the AI to point, click, and navigate applications on behalf of users. Available as a research preview for Pro and Max subscribers, the update marks a significant step in the development of AI agents capable of operating within real desktop environments. Key Developments i What it does for you Instead of giving Claude step-by-step instructions, you can now delegate entire tasks. Claude observes your screen independently, operates the cursor, clicks, and navigates applications — handling everything from file management to complex multi-step software workflows while you focus elsewhere. iiBuilt on Dispatch The update builds on Dispatch, released March 17, which allows users to assign tasks remotely from their phone to a persistent Claude session running on their Mac. Send a task from anywhere — Claude executes it on your desktop. iii Early results Initial testing by MacStories showed mixed outcomes. Simple tasks such as summarizing emails performed well. Browser and Terminal actions failed more frequently — expected behavior for a feature in early research preview stage. "The technology holds clear potential to reshape how we use computers — but requires further refinement before it can reliably support daily use." The release signals a broader shift in how AI assistants interact with our tools. Rather than chatting about tasks, Claude is beginning to execute them. For professionals and builders, that distinction matters. Skip the theory watch what ships next.

Hey everyone 👋 I’m excited to officially open up Falcon Builder. Falcon is a no-code platform for building and running AI agents and workflows — with real triggers, real integrations, and real execution. You can: • Build workflows visually • Trigger via webhook, schedule, or email • Use OpenAI, Claude, Gemini, or Ollama • Connect to databases and APIs • Monitor executions with full logs • Scale with built-in billing and usage limits This is not a toy or demo builder — it’s designed for production automation. It's still early, and I'm continuing to develop and add new nodes, triggers, etc. You can sign up here: 👉 https://www.falconbuilder.dev I’d love feedback from builders, agencies, and operators who are deploying AI in the real world.

Im curious to know your profession / industry? Use this post to find friends in the same space!