01What is GPT-6 Astra second brain and how it works

GPT-6 Astra second brain is a system that gives your AI agents a shared memory. It stores everything you work on, so your agents can give you more personalized answers without you repeating yourself.

In this episode, Julian Goldie shows how he built this system using Obsidian, a free open source tool that holds folders and files. GPT-6 Astra sets up the memory galaxy, and your AI agents update it automatically every time you use them. The agents read the memory when you ask questions, so they know your recent projects, competitors, and workflows.

This makes your AI agents more efficient. You spend less time prompting because they already understand your life. The system also creates a positive feedback loop: the more you use your agents, the better the memory gets, and the better the agents perform.

02Main features of the second brain system

The second brain has several key features that make it stand out.

  • Memory galaxy: a 3D graph where each star is a memory. Bright stars in the middle are the most recent and most connected.
  • Automatic updates: AI agents like Hermes, Codex, and Cal write new memories for you, so you don't have to do it.
  • Creator mode: turn your memories into new content, like SEO articles, apps, or social media posts, with one click.
  • Model-agnostic: works with any AI model, even future ones like GPT-7 or Fable 9, because it's just markdown files.
  • Free and open source: Obsidian costs nothing to use, and the whole system is built on standard folders.

Julian points out that the system is beautiful and fun to explore. The graph view shows connections between memories, making it easy to see what you've worked on recently and what relates to it.

03How to set up the second brain with GPT-6 Astra

  1. Download Obsidian for free from the official site. It's a regular folder-based app, so you don't need any special setup.
  2. Open GPT-6 Astra and give it a prompt: "Build this Obsidian system and organize it for me based on my recent memories."
  3. Tell the agent to update Obsidian every time you use it. Say: "Every time I use you, make sure you update Obsidian with the workflows you worked on, in a straight-to-the-point, easy-to-understand way."
  4. Let the agent create the folder structure and memory galaxy for you. You don't need to build it from scratch.
  5. If you want the full system with all agents connected, Julian offers the Agent OS inside the AI Profit Boardroom, but you can start with just the prompts above.

Once set up, the system runs itself. Your agents update it in the background, so you don't have to spend time maintaining it.

04How to use the second brain for content and apps

After the system is running, using it is simple. In the memory galaxy, you see all your recent memories on the left. You can select one or multiple memories to work with.

For example, Julian picks a memory about motion design in Astra 6. Then he chooses an output, like an SEO article or an app. He writes a brief, such as "Write a practical SEO article from these memories, pick one clear search intent, explain the workflow, and separate the fast from suggestions." Then he clicks "Create with Codex," and the agent builds the content.

He also shows a social media generator. It creates hooks, a seven-part thread, and LinkedIn posts based on his recent work. Everything is personalized because the system uses his actual memories.

Even if an agent didn't work on a memory, it can still read and use it. Julian asks Hermes for automation ideas, and it looks at all memories, not just its own, to suggest five things to automate.

05Results and comparisons shown in the episode

Julian compares plain Obsidian with the memory galaxy. Plain Obsidian is just folders and files—nothing exciting. But when plugged into this system, it becomes a beautiful 3D graph with constellations and bright stars.

He demonstrates a real example: an app built for motion design in Astra 6. The agent saw from his memories that he had been working on motion design recently, so it created a dedicated app. The app lets you enter a title, message, and video duration, then hit generate.

It works without any manual updates.

He also shows that the system works across multiple AI agents: Codex, Hermes, Openglore, Kimikode, GLM 5.2, and OpenClaw. All of them update the same memory system, and automations like Hermes Oracle refresh memories for you as well.

06Tips, limits, and mistakes to avoid

Julian shares a few important tips from his experience.

  • Don't overcomplicate the automation. Only automate things that make sense; otherwise, you waste time.
  • Keep memories straight to the point. If they're too long, other AI agents may get overwhelmed and not use them effectively.
  • Make sure your agents update the memory with relevant information only. Irrelevant memories clutter the system.
  • Don't worry about new models. Since the system is model-agnostic, you can plug in any future AI agent without rebuilding.
  • Use the templates you create, like the SEO brief, to save time. Julian uses a simple prompt that gives the agent a clear job.

The system has no real limits—it's free and doesn't require maintenance, but you do need to give good prompts initially so the agent organizes things well.

07Who the second brain is for and why it matters

This system is for anyone who uses AI agents for content, automation, or business growth. If you want more personalized answers and less time spent on prompting, this is for you.

It matters because AI models change constantly. A second brain built on Obsidian works with every model, so you don't lose your setup when new AI comes out. It also turns your past work into future content, which helps you grow faster.

Julian uses it for his own business to create SEO articles, apps, and social posts automatically. The system saves him hours of work each week. You can start small with just the prompts he shared, or get the full Agent OS for a more complete setup.