01What Agent OS Is and How It Works

Agent OS is a system that connects all your AI agents, tools, and command-line interfaces (CLIs) in one place and plugs them into a memory system. You can build custom workflows, automate tasks, and manage everything from a single dashboard called mission control.

In this episode, Julian Goldie shows his own Agent OS setup, which includes a voice-activated assistant called Hermes, a Cloud CLI, free cloud code, a local engine for building tools, and a powerful memory system. The system lets you automate SEO content, generate videos with one click, and even create a Kanban board for agent tasks. It is designed to be lightweight and customizable, so you can run it on a laptop or a powerful desktop depending on your needs.

02Main Features of Agent OS

Agent OS comes with several built-in features that make it a complete operating system for AI agents. Here are the main ones Julian demonstrates:

  • Voice-activated Hermes – you can speak with Hermes in real time for hands-free control.
  • Cloud CLI – run command-line tasks from the cloud without local setup.
  • Free cloud code – a free version of Claude for users who don't have a subscription.
  • Local engine – build and run custom tools on your own machine.
  • Memory system (Obsidian) – organize clients and projects as separate stars in an Obsidian Galaxy, so agents can pull context for content generation.
  • SEO automation – plugin keywords and case studies to generate and deploy SEO-optimized content to websites.
  • Video agent pipeline – automate video creation with a single click.
  • NotebookLM integration via MCP – connect NotebookLM to chat with notebooks, do deep research, and generate slides, infographics, and reports.

03How to Set Up Agent OS Step by Step

Setting up Agent OS is straightforward. Julian explains that the system is lightweight, so you don't need a powerful machine unless you plan to run local AI models. Follow these steps to get started:

  1. Choose your hardware – a MacBook Air or any laptop works fine if you only use APIs and CLIs. For local models, use an RTX 590 or DGX Spark.
  2. Pick your operating system – Julian recommends Mac for simplicity, but Windows native (not WSL) and Linux also work well. VPS setups are fine too.
  3. Download the Agent OS zip file – get it from the AI Profit Boardroom classroom, where you also get a full guide and video tutorial.
  4. Install the system – follow the tutorial to set up mission control, connect your memory (Obsidian), and add your first agents and CLIs.
  5. Configure your memory – create separate project folders (stars) in Obsidian for each client or business you manage.
  6. Connect APIs and models – use free APIs from OpenRouter (like Entu or Qwen 3.7 Flash) or paid frontier models like Claude Opus 4.8 or GLM 5.2. You can test models side-by-side at goldiebench.com.
  7. Customize your workflows – hide, delete, or add modules like the SEO agent, video pipeline, or NotebookLM MCP. You can even build your own UI as many community members have done.

04How to Use Agent OS: Examples from the Episode

Julian walks through several real-world uses of Agent OS. Here are the main examples:

  • SEO content automation – from the mission control dashboard, you plug in a keyword and a case study, then deploy the content directly to your website. Julian shows an example of a nicely formatted, SEO-optimized post that ranks inside Google AI Overviews and on the first page.
  • Video agent pipeline – with one click, you can automate video creation. This helps market your business and get more leads.
  • Obsidian memory for client management – Julian organizes each client as a separate star in Obsidian. When he asks the agent to generate content for a client, it pulls the relevant information from that star.
  • NotebookLM integration – using an MCP, you connect NotebookLM inside Agent OS. You can chat with notebooks, do deep research, and generate audio overviews, slide decks, quizzes, data tables, mind maps, infographics, flashcards, and reports. All outputs are saved in one place.
  • Custom builds – community members like Marlon and Jose have customized the UI and added their own style. Abhishek connected Google Search Console, keyword research, competitor analysis, and deployment to WordPress and Shopify.

05Results and Comparisons from Julian's Testing

Julian shares specific results from using Agent OS for SEO. One community member, Gabrielle, is ranking inside AI for his keyword. Julian shows his own SEO content ranking inside the first sentence of Google AI Overviews and also on the first page of Google.

For model comparison, Julian created GoldieBench at goldiebench.com. He tested many models and found the top two frontier models to be Claude Opus 4.8 and GLM 5.2. Others that performed well include Grok, Grok build, Qwen 3.7, and Kimi K2.7.

The site lets you compare models side by side on tasks like content creation and see overall scores.

Julian also compares Agent OS to n8n. He explains that n8n is good for beginners to learn automation basics, but Agent OS is for advanced users who want a complete system with memory, custom workflows, and direct integration with tools like Obsidian and NotebookLM.

06Tips, Limits, and Mistakes to Avoid

Julian gives several important warnings based on his own experience and community feedback. First, never connect your personal email to Agent OS. Use a sandboxed email address that only the agents can access.

This avoids accidental deletions or security issues, similar to problems seen with OpenClaw.

Second, don't assume you need expensive hardware. The Agent OS system itself is very lightweight. You can run it on a MacBook Air or any laptop unless you want to run local AI models.

For local models, an RTX 590 or DGX Spark is best; a Mac Studio is not good unless it's the highest spec version.

Third, choose your setup carefully. On Windows, go native – not WSL – because WSL adds complexity. If you travel a lot, a MacBook Air is fine.

You can also use a VPS and give your agent OS mobile and desktop access.

Finally, customize and sandbox. You can hide or delete parts of the system you don't need. Build your own version and style, as many community members have done.

Use a free API like Entu or models from OpenRouter if you don't want to pay for subscriptions.

07Who Agent OS Is For and Why It Matters

Agent OS is for anyone who wants to connect multiple AI agents, tools, and workflows in one place. It is especially useful for service-based businesses, agencies, and SEO professionals. Julian highlights three key use cases: automating video creation to get more leads, generating SEO content that ranks in Google and AI overviews, and organizing client projects using Obsidian memory.

It also matters because you can customize everything. You don't have to use the default setup – you can build your own UI, add your own agents, and tweak the system to fit your exact needs. The community shows examples of custom versions that include Google Search Console integration, Shopify deployment, and competitor analysis.

If you want support, Julian answers questions every day in the AI Profit Boardroom community and provides daily updates to the system.

08How to Get Agent OS and Join the Community

To get Julian's Agent OS setup, you need access to the AI Profit Boardroom. Go to the classroom and then to the Agent OS section. There you get the zip file, a full guide, a video tutorial, and daily updates when new features are added.

You also get four weekly coaching calls where you can share your screen and ask questions. The community has over 90 people online at any time, ready to help. You can post questions and Julian answers them every day in a video tutorial.

There is also a map feature to meet local people who are building with AI agents just like you.

To join, go to aiprofitboardroom.com or click the link in the description. Julian notes that he and his agents answer questions fast – one community member got a full video tutorial just two hours after asking a question.