01What Is an Agent Operating System?
An agent operating system is a single dashboard that runs your voice agents, video agents, goals, and memory system in one place. It makes your AI agents up to 100 times more powerful because everything is connected and automated. You can set a goal, walk away, and wake up to finished work.
Julian Goldie built his own agent operating system over 100 hours and shares the process. He says the best part is that every part of it is free to build yourself if you use the right tools. The system turns a messy pile of AI tools into one organized workflow.
Instead of using AI like a calculator with one question and one answer, you run AI like an operating system. That means memory, agents that build overnight, and every output on one dashboard. This is what the episode is all about.
02Five Power Moves in the Goldie Build Anything Stack
The Goldie build anything stack has five parts that work together. You do not have to use all five from day one. Each one can stand alone, but the magic happens when they link.
- Voice lets you talk to your AI agents instead of typing. Julian uses OpenClaw to have a phone-like conversation.
- Goals let you set a long-horizon objective and walk away. The agent runs autonomously for up to 20 turns.
- Video turns a single prompt into an AI avatar video using HyperFrames, a free open-source project.
- Search gives you a formatted Markdown guide from a query, using Twitter search linked to OpenCore for latest news.
- Memory saves everything you create in an Obsidian vault so it is searchable forever.
These five parts form what Julian calls the build anything stack. When you connect them, you get a system, not a tool stack. For example, voice can talk to goals, goals can create websites, and a SEO deployment system can save them.
03How to Get Started Building Your Own System
Building an agent operating system is easier than you think. Julian says you can set it up in about an hour once you know the steps. Here is how to start.
- Open Hermes and ask it to build you an agent operating system. Explain what you want, and it can create it for free.
- Set up the dashboard. Use a local directory so all files are stored on your computer.
- Add the voice agent. Go to OpenClaw, then studio, then talk. Click and start speaking.
- Set up goal mode inside Hermes. Use the goal mode to run autonomous tasks in a sandboxed directory.
- Create the video agent by installing HyperFrame CLI and connecting it to your dashboard.
- Add the search system for research and the memory system using Obsidian for storing outputs.
You do not need to be technical. Julian says he does not code, but he built everything. If he can do it, you can too.
Just start with one part today.
04How to Use the Agent Operating System in Daily Work
Once set up, using the agent operating system is simple. You talk to your agents, set goals, and manage everything from one screen. Julian shows several examples in the episode.
For voice, you open OpenClaw, go to studio, and click talk. You can have a live conversation with your agent just like a phone call. No typing needed.
For goals, you set a goal on Sunday night, and the agent builds a website by Monday morning. It writes pages, creates blogs, and exits once the goal is met.
For video, you type a prompt inside your AI agent, and it generates an avatar video in one click. You can preview all your creations in the workspace. For search, you type a query and get a proper Markdown guide with latest news.
For memory, you ask Hermes to use your Obsidian vault to suggest SEO keywords. It gives personalized ideas without a rebrief.
05Results and Improvements Shown in the Episode
Julian shares results from his own system and from community members. He built the video agent in one morning and the agent teams orchestration Kanban board worked perfectly. People like Dr.
John Meza created their own agent operating systems and got great results.
The system saves a lot of time. For example, the goal mode runs autonomous tasks while you sleep. You wake up to finished work.
The memory system stops losing files. Every voice conversation, search result, and video is saved and organized.
Julian says the more you use the system, the better it gets. Each new part becomes 10 times more useful when connected to others. For example, creating videos and SEO content together becomes powerful.
This positive feedback loop makes the whole system stronger daily.
06Tips, Limits, and Mistakes to Avoid
Here are some tips from the episode to avoid common mistakes. First, do not wait to build. Julian says the window to learn is now, and people who wait fall behind.
He also warns against using AI like a calculator with one-off questions. Build a system instead.
You do not need to build everything at once. Start with one part and add more later. Each part teaches you the next.
Take your time and improve a little each day. Julian also says not to install tools separately. That leads to 10 tabs, many subscriptions, and no shared memory.
Instead, link everything in one dashboard.
If something breaks, it is stored locally, so you can ask your agent to restore the last version. It is also backed up. One limit is that the episode does not show a full technical setup, but Julian offers a zip file and video tutorial in his community.
Finally, be careful with OAuth logins. You can use Hermes with ChatGPT or Grok accounts to generate images and videos.
07Who Should Build an Agent Operating System and Why
An agent operating system is for anyone who wants to get more done with AI. Julian built it because he was tired of juggling many tools and losing files. He wanted one place to run everything.
This system is for people who want results, not just playing around with AI tools.
If you are not technical, you can still build it. Julian is not a coder. The system is simple because it uses Markdown files inside Obsidian.
You can ask Hermes to build it for you, and it is free.
Building this system matters because it changes how you use AI. Instead of typing every message, you talk. Instead of managing agents, you walk away.
Instead of losing outputs, you keep everything. It saves time and makes your AI agents more powerful. Julian says people who build today will look back in six months and see this as the moment they started winning.
