What I Did in the First 10 Days After Launching My Open-Source AI Tool (The Real Story)

Most launch stories you hear are flashy: "Launched on Hacker News, got 1,000 stars overnight."

This isn’t one of those stories.

This is a real one.

In the first 10 days of launching my open-source tool — Kaizen Agent
— I got:

• ⭐ 15 GitHub stars
• 🍴 3 forks
• And 9 of those stars came from my engineering friends I personally messaged

But those early days were incredibly valuable — not because it went viral, but because the feedback I got helped me move forward fast.

---

😅 I almost didn’t launch

To be honest, I was a little hesitant to launch.

The onboarding process wasn’t polished. The tool wasn’t perfect. I thought,

“Should I wait until it feels more complete?”

But I decided to post anyway — just to see what happens.

And that’s when everything started moving.

---

📣 Where I launched

In the first few days, I:

• Posted to Hacker News: https://news.ycombinator.com/submitted?id=yuto_1192

• Shared on Reddit: https://www.reddit.com/r/AIAGENTSNEWS/comments/1lobzw8/tired_of_trial_and_error_to_improve_your_ai_agent/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

• Created a new Twitter account: @yuto_ai_agent

• Sent it to some engineering friends

No major launch strategy — just shipped it and started talking about it.

---

🧠 The feedback that changed everything

After launching, I got a few important messages — from friends and Reddit comments — that really helped.

The key feedback:

“It’s cool, but I didn’t really know how to get started.”

That was 100% valid. My onboarding wasn’t clear. The README was dense. It wasn’t easy to try.

So I paused any further promotion and focused on making the product easier to use.

---

🔧 What I improved

1. Rewrote the README

   • Made it simpler
   • Added a dead-easy example
   • Focused on clarity

2. Published to PyPI

   • So people could run pip install kaizen-agent
   • No more cloning and pip-editing

3. Launched a docs site

   • Documentation here
   • Added a proper walkthrough for YAML format and usage

---

📈 What changed

After improving the onboarding:

• GitHub star conversion rate increased significantly
• Strangers forked it

📊 Screenshots of traction

Here are two screenshots showing the traction from GitHub traffic and stars:

💡 What I learned

• Launch early, even if it’s imperfect
  
  As long as the core function works, feedback is worth more than polish.

• README is your first impression
  
  If people don’t understand it in 10 seconds, they won’t try.

• Ask for feedback
  
  Especially from AI developers working with LLMs or agents — it’s how I found direction.

---

🙏 Final thoughts

If you’re building an AI tool or LLM app, and wondering if it’s “ready” to share — launch it. Just make sure the core thing works.

Ask for feedback. Then improve from there.

If you're curious, here’s the project:

👉 https://github.com/Kaizen-agent/kaizen-agent

And if you work with LLMs or AI agents, I’d love your thoughts or feedback.

Thanks for reading!

— Yuto