Thank you Jake Van Clief and David McDermott for having me, and David Vogel for the Hermes assist. This one is ICM in practice. I open my own Hermes and walk every layer of its memory, live: the built-in memory files, session history, skills and the curator, the memory provider I host myself, and the ICM workspace that holds the job itself. The part I think this room will care about: in my workspace, every folder carries a small AGENTS.md router that points to a CONTEXT.md contract. Hermes loads AGENTS.md on its own. It doesn't load CONTEXT.md, so the router is what makes sure the contract gets read. Small router, real instructions 1 step away. Watch it here: https://www.youtube.com/watch?v=GtROa4PfvY8 ICM is why this system holds together, and Hermes is what walks it. Jordan The read through https://agentsasfolders.ai/blog/ai-agent-memory-hermes/

HERMES WORLD ENGINE Hermes World Engine is a folder-based creative workflow for building consistent characters and worlds, producing approved images, and animating those images. Briefs, specifications, references and review records preserve creative intent across generations. I built this entirely from scratch inside my Hermes Agent with the Desktop App on a Hermes Agent gateway on a VPS. I did not use anyone else's template but future builds will incorporate the ideas from Jake's recently re-made animation build. I've used eleven labs before , but the quality that Jake produced made me a believe it can be done properly. Build status Working prototype—not yet a full studio. Built and exercised: world/character records, image production, approvals, and brief-led image-to-video animation through Hermes → Nous → Seedance 2.5. Proven output: two accepted animated clips. All four workflow tests currently pass. Designed, not implemented: optional intake routing, story foundations, maps, multi-angle reference libraries, storyboards and sequence continuity. Future composition: voice, sound, captions and final assembly using tools such as ElevenLabs and Remotion—after animation assets are accepted.

This might actually be the most powerful thing I've ever created, and I honestly didn't know it would work this way when I did. I've been working on sharing workspace with my agents for a while. My janky setup: open a Google doc, share the link with whichever agents I want, and tell them to use chrome to come meet me to work on something. When that first worked, it lit a spark. The concept was real. So I set out to build a shared workspace for me and my agents that suited me. I went with Infinite Whiteboard as my shared workspace format. I knew it would be good. I knew it would be an unlock. I didn't realize how POWERFUL it would be with Cortana and my other agents driving it! I've attached how simply I asked for this functionality after creating the shared spaces and using them mostly for scribbling ideas.

Please forgive the unbridled enthusiasm! It has been my mission to make the most functional AI system I could possibly imagine. That has never had one shape. It's a feeling. A North Star. Without stepping into overstating anything, this feels like it. This feels like the Autumn's AGI moment! Local voice. Local live avatar. Custom native android app. Custom android launcher. Custom notification widget made more beautiful by sending Cortana ONE TWEET and saying "this would look nice for your noti pill!" The face on my screen can answer from disk when simple, and passes off to the full Cortana at home, then Cortana appears with her message! PLEASE ask any questions! I've been deep in this so long I don't even know where to start anymore!

I gave a talk at my company’s ten year expo this week and wanted to share the recording here. It's a high-level look at some of the ways we're putting AI to work in an automation manufacturer, and I got a lot of nice compliments from folks afterward. I walk through three levels, and each one gives the agent more context than the last. The first is a quote auditor that runs on a checklist. It reviews a 200-line vendor quote against our drawings and order summary and hands back the five lines that actually need my attention. The second is a panel generator that runs on an engineer's rulebook. It turns a pile of components and about a hundred layout rules into a draft electrical panel, taking a job that could eat days down to as little as 15 minutes. The third is a portfolio assistant that runs on the whole company. It sits in on every meeting through a transcription app we built ourselves, so nothing leaves the building, and it connects what it hears across all our projects. That being said, the theme I think everyone here can take away has nothing to do with automation. It's not about the tools you use or keeping up with the latest model. Out of the box, these models give you average output. What makes them valuable is your expertise, written down and handed to an agent. The same message you see me trying to explain in all of my content, highlighted with real examples. Take a look and let me know what you think!