By the end of this guide, you will know exactly how to build your own AI operating system (AIOS) using Claude Code and Cloud Code, even if you have never used these tools before. This article provides a detailed walkthrough of the entire setup process, mindset shifts, frameworks, and practical demonstrations to help you create a powerful AI assistant that can automate and streamline your business operations.
An operating system (OS) is traditionally the layer between you and your computer or phone, managing files, apps, and contacts. With AI integration, an AI operating system adds intelligence on top of this layer, enabling agents that can see, interact with, and remember your business data better than a human can.
Imagine spending an entire workday with just Cloud Code open, accomplishing more than others who switch between multiple apps. This is the power of an AIOS.
The Three M's of AI: Mindset, Method, and Machine
Mindset
- Default Shift: Before any task, ask yourself how AI can do part or all of it. Even 30% automation is a huge productivity gain.
- Function Breakdown: Break your role into small tasks and automate each chunk step-by-step.
- Curiosity Rule: Treat AI as a mentor, not a vending machine. Always ask why and understand the code or process AI uses.
Method
Decide what is worth automating and to what extent. This framework helps prioritize automation efforts.
Machine
The technical aspect of building and running your AIOS, including tools, APIs, and integrations.
The Four C's Framework for AIOS
- Context: What AI knows about you, your team, tools, business, and voice.
- Connections: Data sources AI can access (e.g., CRM, calendar, email).
- Capabilities: What AI can do with the data (e.g., generate reports, send messages).
- Cadence: When AI acts autonomously (e.g., scheduled tasks, routines).
These build on each other sequentially and are essential for a functional AIOS.
Setting Up Your AIOS Project
- Use Visual Studio Code (VS Code) with the Claude Code extension.
- Create a project folder (e.g., AIOS) and open it in VS Code.
- Clone the AIOS GitHub repository provided in the free school community.
Onboarding Your AIOS
- Run the onboard skill to answer questions about your business, priorities, and communication style.
- Provide detailed context to help AI understand your needs.
Connecting Data Sources
- Identify key tools: revenue systems (Stripe, QuickBooks), customer platforms (CRM, YouTube), calendar (Google Workspace), communication (email, Slack, ClickUp), tasks, meetings (Fireflies), and knowledge bases.
- Connect these tools via APIs or MCP servers, providing API keys securely in environment files.
- Use separate API keys for AIOS accounts to control permissions and security.
Building and Using Skills
What Are Skills?
Skills are reusable recipes or SOPs that automate specific tasks, such as writing LinkedIn posts or creating slide decks.
Anatomy of a Skill
- skill.md: Markdown file with name, description, step-by-step instructions, and references.
- Supporting files: Scripts, references, or assets linked to the skill.
Creating Skills
- Use the skill builder to define the problem, goal, workflow, triggers, and output.
- Skills can be invoked via slash commands or natural language.
- Skills improve over time through feedback and iteration.
Testing and Iteration
- Watch the AI perform the skill and provide feedback.
- Fix issues like wrong steps, missing context, or repeated mistakes by editing the skill.md or adding references.
- Use rules to add guardrails and constraints.
Automating with Routines and Loops
Routines
- Scheduled tasks that run on the cloud or locally.
- Can be triggered by time, API calls, or GitHub events.
- Support environment variables for API keys and permissions.
Loops
- Recurring tasks that run within a session for up to 3 days.
- Useful for frequent checks like monitoring emails or deployments.
- Can be created, listed, and deleted via natural language commands.
Differences
- Routines run independently of your machine being on.
- Loops require the session to be open and have a 3-day expiry.
Leveraging Google Workspace CLI (GWS CLI)
- GWS CLI provides access to Gmail, Drive, Docs, Sheets, Calendar, and more through a single interface.
- Enables multi-step workflows and automation recipes.
- Integrate GWS CLI with Cloud Code to automate document creation, data extraction, and organization.
Building a Knowledge Base with LLM Wiki
- Inspired by Andrej Karpathy's LLM wiki concept.
- Organize raw data (e.g., PDFs, transcripts) into markdown files.
- Use Obsidian to visualize relationships between documents.
- AI can query this knowledge base efficiently, reducing token usage by 95% compared to traditional semantic search.
Visualizing and Managing Your AIOS
- Use Obsidian or similar tools to visualize your knowledge graph.
- Maintain indexes and logs for your knowledge base.
- Continuously ingest new data and run linting to ensure data quality.
Success Criteria for Your AIOS
- Your team can interact with the AIOS instead of bothering you.
- You reduce context switching and tab clutter by working mostly within Cloud Code.
- Knowledge is externalized from your head into the AIOS, reducing stress and improving memory.
Final Thoughts
Building an AI operating system is a journey of continuous improvement. Start with your personal AIOS, connect your key tools, build reusable skills, and automate repetitive tasks. Use routines and loops to maintain cadence, and leverage knowledge bases for efficient information retrieval. With dedication, your AIOS will become an indispensable assistant, boosting your productivity and preparing your business for the AI-driven future.
For additional resources, templates, and community support, join the free school community linked in the description of the original video.
This guide encapsulates the detailed process and mindset shared by Nate, an AI automation expert, in his comprehensive course on building and selling Claude Code operating systems.
