Hacker News
This article by Ron Forbes on Hacker News AI provides a tutorial on how to build a personal AI assistant using Claude Code, emphasizing automation of low-stakes tasks to enhance productivity without requiring extensive coding knowledge.
AI 生成摘要
這篇由 Ron Forbes 在 Hacker News AI 發表的文章，提供了一個利用 Claude Code 建構個人 AI 助理的教學，強調自動化低風險任務以提升生產力，且無需深厚的程式設計知識。
How to automate the low-stakes tasks consuming your time so you can focus on the high-stakes decisions that actually matter
TL;DR: Build your own AI assistant in 15 minutes. No coding. I'll show you exactly how I went from zero to 2,000+ organized notes.
We finally have AI superpowers. You can generate code, analyze data, automate workflows. The future is here.
So why do you feel more overwhelmed than empowered?
Every week there's a new AI tool, a new technique, a new "game-changer" you're supposed to learn. Meanwhile, you're hearing that AI is coming for your job. The anxiety is real.
Here's how I think things might play out: AI isn't going to take your job. The AI-native person who augments their productivity with AI is going to take your job.
And that person wasn't born understanding neural networks or transformer architectures. That person started learning just a few months ago by building something simple. Not by mastering every AI tool. Not by becoming an engineer overnight. They started by building their very own personal assistant.
In case you're new here, I'm Ron, a product manager at Meta Quest. I'm a PM who started learning how to vibe code with Claude Code, but I quickly realized you could use Claude Code for much more than just software development. Over the past month, I've built RonOS: a personal operating system with linked notes, automated meeting summaries, daily priorities, and insight synthesis. It's also my writing and thinking partner. It helps me go from idea to outline to draft to published content much faster.
I didn't set out to build this. I just started automating one annoying thing. Then another. Then another. It compounded.
In this guide, I'll show you how to build YouOS, your own personal operating system. Not a copy of mine. Yours. One that works the way your weird brain works.
No engineering background required. Just 15 minutes and the willingness to start.
For the first time in history, you can build software that works exactly how YOU work. Not how Asana thinks you should manage tasks. Not how Empower thinks you should view your budget.
Software that fits your idiosyncratic, weird, perfectly-logical-to-you way of thinking.
Real examples from my life:
Daily brief that connects my thinking: Every morning I get a brief on my phone with insights from recent journaling, key themes from my work, my three priorities for the day, and how my current thinking connects to my broader goals.
Budget tracking that reflects my values: Instead of categories like "Dining Out," I see how my spending aligns with what I actually value: health, learning, growth, experiences, and relationships. That's insight I can act on.
Meeting intelligence for back-to-back days: Claude Code processes my meeting notes in real-time. At the end of the day, I have a clean, prioritized list of what needs action.
The shift:
Traditional software forces you into THEIR boxes. Personal software fits YOUR boxes. You build exactly what you need, iterate as you evolve, and own your data completely.
But here's the requirement: You have to take initiative. The software won't build itself.
Here's what shocked me:
People in completely non-technical roles are building with AI TODAY. Teachers, writers, therapists, lawyers, small business owners, stay-at-home parents managing household chaos.
I'm a PM who hasn't written production code in years. I build stuff by vibing with Claude Code, describing what I want and iterating.
If I can do it, you can do it.
Hint: We might have a guest post soon from someone in a very non-technical role who's unlocking their own superpowers through AI. Stay tuned.
The barrier isn't technical knowledge. The barrier is believing you can do it and taking action.
When I tell people I built RonOS with AI, they ask: "But what should I build first?"
I get where they're coming from. You CAN build anything now, which paradoxically makes it nearly impossible to decide WHAT to build.
The solution: Build your personal assistant.
An assistant that handles the tedious stuff you're ALREADY doing:
Why this approach works:
You solve your own problems first: Your personal assistant reveals what workflows are actually valuable to automate.
Baby steps with immediate value: You automate the things you're already doing. Every automation saves time immediately.
Compound learning: Week 1 you learn how Claude Code works with files. Week 2 you learn to synthesize new insights from your data. By month 2 you're building custom tools that seemed impossible before.
My RonOS journey:
I didn't plan to build a personal operating system. I just felt like I was running from meeting to meeting, drowning in notes. So I asked Claude Code: "Help me organize today's meeting notes into something useful."
That worked. So I tried: "Extract the action items and who owns them."
That worked too. So I kept going. Then I automated my daily priorities. Then my insight synthesis from years of hastily written notes. Then my budget tracking.
A few weeks later, I had organized notes, automated workflows, and a system that's already helping me live more intentionally. It happened through compound automation, not a master plan.
Most people miss the critical distinction between Claude Chat and Claude Code.
Claude Chat: Transactional. Best for quick questions, web research, one-shot content drafting. You manually copy/paste, upload files, recreate context every session. Like texting an advisor.
Claude Code: Autonomous agent with file system access. Best for multi-step workflows, file operations, rapid iteration. Maintains context across sessions, reads/writes your files, executes tasks independently. Like having a chief of staff with access to your office and systems to be your strategic partner, force multiplier, and integrator of workstreams.
What takes 30 minutes of copy-paste in Claude Chat takes 2 minutes of delegation in Claude Code.
The difference between asking for help and delegating entirely
Here's the delegation insight:
When I tell Claude Code to process my meetings, I'm not asking it to help me do the work. I'm delegating the work entirely.
I think this is a skill that will define career success in the AI era. The future belongs to people who can have a clear vision, communicate it in bite-sized pieces to an agent, validate what the agent produces, and orchestrate several agents at once.
Sound familiar? This is product management. And it's a learnable skill.
What you need:
Before we set up your personal operating system, let's see what Claude Code can actually do. We're going to clean up that messy Downloads folder you've been ignoring.
Install Claude Desktop (simplest option):
Alternative: VS Code + Claude Code Extension:
Now run your first automation:
Tell Claude Code:
"Analyze what's in this folder and organize the files into logical subfolders based on file type and content. Before you move anything, show me your proposed organization plan and let me approve it."
Watch Claude work. It will:
What just happened: You delegated a task you've been putting off for months. Claude read your messy folder, made sense of it, and organized it for you, all while asking permission before making changes.
This is the core loop: describe what you want → let Claude propose a solution → approve and execute.
Boom. You're an AI wizard.
How did that feel? Reply and tell me what Claude found as it cleaned up your Downloads folder.
Now let's build a system to apply this power to the rest of your life.
On Mac:
On Windows:
Your folder structure should look like this:
Obsidian is where you'll write and organize your notes. It's a beautiful, fast note-taking app that works with plain text files.
Install Obsidian:
Understanding Markdown:
All your notes will be markdown files (.md). Markdown is simple text formatting that looks good in Obsidian. Here's everything you need to know:
To create a new note in Obsidian:
That's it. Don't overthink it — you'll learn more as you go.
Now point Claude Code at your new YourNameOS folder. You have three options:
Option A: Claude Desktop (simplest)
Option B: VS Code + Claude Code Extension (recommended for power users)
Option C: Obsidian + Terminal Plugin (stay in one app)
All three options give Claude Code the same capabilities. Pick based on where you prefer to work:
Problem it solves:
When you're in back-to-back meetings all day, you can't stop between each one to process notes. By 6pm, you might have a dozen sets of notes with no clarity on what needs action.
The automation:
Step 1: Create your structure
Step 2: Process meetings at end of day
After each meeting, dump your notes into /meetings/raw. Don't organize. Just brain dump. You can even put them all in one file if you like.
At the end of the day, tell Claude Code:
Time saved: 30-45 minutes of manual processing per day. That's 10-12 hours per month now spent on actual strategic thinking instead of administrative cleanup. This gets even easier with AI meeting note taking tools like Zoom, Granola, and Otter. Now you can focus fully on the discussion and know that AI will extract the outcomes and action items into your knowledge base.
Once you have meeting intelligence working, consider adding a personal layer.
The concept: Create morning and evening routines that take 2-3 minutes each but give you clarity for the whole day.
Step 1: Create your commands folder
In your YourNameOS folder, create this structure:
Pro tip: you may need to show hidden folders (those starting with a dot) in order to see .claude. On Mac: In Finder, press Cmd+Shift+..
On Windows: In File Explorer, click View → Show → Hidden items.
Step 2: Create your morning command
Create a file called morning.md in the .claude/commands/ folder with this content:
Step 3: Create your evening command
Create a file called evening.md in the .claude/commands/ folder with this content:
Step 4: Run your commands
In Claude Code, simply type:
That's it. Claude reads your command file and follows the instructions.
The key: These are YOUR templates. Edit them to match how you think. Maybe you want to track exercise, creative projects, or habit streaks. Make it yours.
The compounding insight: After 30 days, you have 30 daily notes showing patterns in what you avoid, when you have energy, and what types of days feel successful. After a few days, you can start asking Claude to analyze your patterns and give you real data about how you actually work.
My results after 30 days:
The Critical mindset shift: Do NOT try to build everything on day one.
Week 1: Pick ONE thing (meeting summaries, daily notes, task extraction). Get it working. Use it for a week.
Week 2: Add ONE more thing that builds on the first. Link meeting summaries to tasks. Create weekly reviews.
Month 2: Now you have enough data to get insights. Ask Claude to analyze your meeting patterns, find what you consistently avoid, compare stated priorities vs completed tasks. Build custom tools and integrations you couldn't have imagined on day 1.
My Actual RonOS Timeline:
Here's exactly how my system evolved, pulled from my git history:
I didn't plan any of this out in advance. I just kept solving the next most annoying problem, freeing up mental capacity to focus on what really counts.
A lot of my day is consumed by low-stakes tasks: reformatting notes, copying action items, searching for that thing someone said three weeks ago.
Sadly, the high-stakes decisions often get whatever energy is left at 6pm: Should we prioritize this feature? How do we balance user experience with technical constraints? How do I show up with presence for the people I care about? Am I living deliberately or just responding to whatever comes at you?
It turns out I had the equation backwards.
Low-stakes tasks should be automated. High-stakes decisions should get your best energy.
That's what I'm loving about building RonOS. It automates the low-stakes noise so I can devote my energy to what matters most.
Not tomorrow. Not next week. Today.
Choose one workflow:
Set up in 15 minutes. Follow the setup guide above. Get your first automation working. Don't overthink it.
Ship it messy. Your first version will be rough, and that's okay. The goal is to start the compounding process.
Remember:
We're entering an era where the limiting factor isn't AI capability — it's your willingness to take initiative.
Claude Code gives you superpowers. What you do with them is up to you.
AI isn't going to take your job. But the person who learns to delegate to AI, orchestrate agents, and build custom solutions for their unique needs? They might.
I built RonOS one automation at a time. You can build YouOS the same way.
The question isn't "Can AI help me?"
The question is "What can I delegate to AI today?"
I'm documenting my entire RonOS journey: the wins, the failures, the "wait that actually worked?" moments. And I'm sharing everything I learn about building with AI as a non-engineer.
Join readers who are building their own YouOS systems:
👉 Subscribe to The Degenerate
No fluff. No hype. Just a PM building in public and sharing what actually works. I'm dropping more advanced experiments with RonOS as my personal assistant. Subscribers get them first.
See you in your inbox.
I'm building my AI productivity system in public and documenting everything. Follow along for weekly experiments with Claude Code, Obsidian, and whatever I'm building next.
Product Manager at Meta Quest. I write about AI productivity, personal knowledge management, and building in public.
© 2026 Ron Forbes. All rights reserved.