01What Is an Agent OS and How It Works

An agent OS is a system where you put all your AI agents in one place. It lets you run them and build custom workflows. The best part is you can build your own agent OS free.

You do not need to pay for expensive APIs. In this episode, Julian Goldie shows you exactly how.

The system works by using a free open source harness as the base. Then you plug in a free model to act as the brain. You use prompts to tell the agent what to build.

The agent creates the OS step by step. You can keep improving it over time.

02The Three-Step System for a Free Agent OS

Julian breaks the process into three simple steps. First, you pick your free harness and model. Second, you start building with prompts.

Third, you add free models and customize the system. This works with any free harness like Deep Sea Harness or Hermes Agent.

  • Step 1: Choose a free harness like Deep Sea Harness or Hermes Agent.
  • Step 2: Pick a free model through Open Code or Open Router.
  • Step 3: Build your agent OS and then keep improving it.

03Choosing the Right Free Harness and Model

To start, you need a free harness. Julian recommends Deep Sea Harness or Hermes Agent. Both are free and open source.

They let you build cool things without paying anything.

For the brain of your agent OS, you can use Open Code. Open Code works inside both Deep Sea Harness and Hermes Agent. You can also use Open Router to find free models.

Just type "free" in Open Router to see options like Light Link 3.0 Flash and LFM 2.5. In Open Code, you can also type "free" to see models like MIMO V2.5 Free and MuseSpark 1.3 Free.

04How to Build Your Agent OS Step by Step

  1. Start a new session in Open Code.
  2. Select a free model, like MIMO V2.5 Free.
  3. Type a prompt asking the agent to build a mission control dashboard for your AI agents.
  4. Let the agent build the dashboard step by step.
  5. Once built, open the local host to see your basic agent OS.

Julian used a prompt like: "Can you build out a mission control dashboard where I can see Open Code running? And I can also build beautiful automations inside there. It should just be a web app that's hosted locally where we can build out our AI agents and have a mission control for all our AI agents." The agent then starts building.

05Customizing Your Agent OS with Free Models

After your basic agent OS is built, you can add more features. Julian shows how to add custom workflows, like an ad builder tool. To do this, you go back to the agent that built the OS and ask it to create the feature.

  • Test the new feature inside your agent OS.
  • If it does not work, give feedback to the agent.
  • Ask the agent to fix what is broken.
  • Keep iterating until it works.

Julian mentions building a web design OS inside the agent OS. It did not work perfectly the first time. He went back and forth with the agent until it worked.

This shows the importance of testing and iteration.

06Results and Comparisons from the Episode

Julian shows that you can build a mission control dashboard for free. He demonstrates that the free models are good enough to build real tools. The example in the episode shows a fully built agent OS with a video editor, memory galaxy, and SEO agent.

He also explains that you do not need NA10 anymore. He says he uses NA10 times zero in his workflows. NA10 requires customization, configuration, and custom APIs.

With the free agent OS, you have all agents and automations in one place, without the pain of NA10.

07Tips, Limits, and Mistakes to Avoid

Julian shares some important tips. First, the initial agent OS will be basic. It will not look as good as the final version.

Do not expect perfection right away.

  • Test everything: Always test new features inside your agent OS.
  • Give feedback: If something is wrong, tell the agent what to improve.
  • Iterate: Go back and forth with the agent until it works.

A common mistake is thinking you need to pay for APIs. But you can use free models and harnesses. Also, do not be afraid to ask the agent to fix things.

The system improves step by step.

08Who Should Use a Free Agent OS and Why

Anyone who wants to build AI agents without cost should use this system. It is good for business owners and personal projects. You can have all your agents in one place and build custom workflows easily.

The free agent OS matters because it removes the barrier of cost. You can experiment and build powerful tools without paying for APIs. Julian also mentions that you can get the whole system inside the AI Profit Boardroom, but the core method is free.

This makes AI agent building accessible to everyone.