01What Is Agent OS AI Tools?
Agent OS AI tools are a set of AI-powered features that Julian Goldie builds inside Claude, the AI model from Anthropic. These tools help him create music, run benchmark tests, and compare AI models, all from one system. He updates the agent OS every day by adding new features based on community feedback.
The agent OS is part of his AI Profit Board community. It is designed to be fun, easy to use, and useful for building AI projects. For example, you can set up a blank slate with Hermes, use loop engineering for complex tasks, or plug in new research updates from NotebookLM.
The goal is to have a flexible system that can handle many tasks and improve over time.
02Key Features of Agent OS AI Tools
In this episode, Julian shows several features that make the agent OS powerful. First, it includes Fusion, an API that combines five different AI models. A judge inside the API picks the best answers, creating a smarter result.
Second, it has GoldieBench, a leaderboard where Claude tracks and judges model outputs from live streams. This helps you see which AI performs best on real tasks.
- Hermes integration: Set up blank slate tasks quickly.
- Loop engineering: Handle complex workflows with ease.
- NotebookLM research updates: Add the latest research directly.
- Fusion: Combine models for better answers (supposed to beat Fable 5).
- Automated guides: Claude creates detailed diagrams for each new feature.
03How to Set Up Agent OS AI Tools
Setting up your own agent OS requires you to use Claude as your base. Julian builds his system inside Claude and adds new features daily. Here are the steps he follows:
- Open Claude and create a new project or workspace.
- Define the core tasks you want the agent OS to handle, like music generation or benchmarking.
- Integrate external tools like Fusion, NotebookLM, or Hermes via APIs.
- Set up a leaderboard system (like GoldieBench) where Claude can track and judge outputs.
- Test your system with small tasks first, then expand.
- Update your agent OS regularly with new features based on what you learn.
04How to Use Agent OS AI Tools in Practice
Julian uses the agent OS in his daily live streams. For example, he plays AI-generated music, usually drum and bass, while he works. He creates this music using the agent OS.
He also answers community questions and builds new features in real time.
One example is benchmarking: he runs tests with Fusion and other models, then Claude judges the results. He shows how the same game, Nordic Crypt, was built by different models with different results. He compares them to see which model is best.
Julian also creates guides with diagrams for each new feature, making it easy for others to follow.
05Results and Comparisons from the Episode
Julian shares several results from his testing. First, Fusion leads 7 to 1 versus GLM, 6 to 1 versus Grok, and 4 to 1 versus Opus 4.8. He says Fusion is creating some of the best stuff so far, especially on the Nordic Crypt game.
Another surprising result: his avatar videos outperform his human videos by 4 to 1. Despite negative comments, avatar videos have grown his channel dramatically. He notes that his AI Profit Board community is top five all time in its category, making $173k MR.
The community has over 186 pages of testimonials.
06Tips, Limits, and Mistakes to Avoid
Julian gives practical advice. First, don't ignore metrics like views—they matter for growth, even if some people say they don't. Second, be open to using avatar videos because they perform better, despite criticism.
Third, update your AI tools regularly to stay current, like adding the latest NotebookLM research update.
A common mistake is focusing only on human content, which can slow growth. Julian also warns that some tools, like the website builder he mentions, can be hard to log into, so you need to rely on Claude to create content. Finally, always test different models to see which one works best for your specific project.
07Who Should Use Agent OS AI Tools and Why
Agent OS AI tools are for anyone building with AI, especially business owners and creators. Julian's community has 3,700 members who use the agent OS to build projects, benchmark models, and grow their channels. It's also for people who want to connect with others and learn AI in a practical way.
Why it matters: AI is moving fast, and having a flexible system like agent OS helps you stay ahead. You can improve your content, make better decisions with benchmarks, and build a community that grows with you. Even if you're a beginner, you can start small and add features over time.
