•
projectsAIentrepreneurship

4 Personal Projects in 40 Days

Over 40 days of weekends and spare time, I shipped 4 personal projects with AI assistance. Here's what I built, why, and what I learned.

Over 40 days of weekends and spare time, I shipped 4 personal projects using OpenClaw as my AI coding companion. Here's the breakdown: what I built, the reasoning behind each, and lessons learned along the way.

The Four Projects

1. Game Strat Port — Mobile Game Guides & Tools

An aggregation site for mobile game guides, starting with Genshin Impact and Zenless Zone Zero.

The thesis: As AI automates more jobs, people will have more leisure time for mobile games and short-form video. Getting in early on this trend seemed worthwhile.

Revenue model: SEO for organic traffic initially, then display ads and affiliate marketing. Eventually, premium membership for advanced tools (~$2.99/month).

2. LLMPick — LLM Reviews & Recommendations

Born from personal frustration—new models drop constantly, benchmarks are confusing, and it's hard to know what's actually good. So I built a site that aggregates real user experiences from Reddit, forums, and social platforms into comprehensive ratings and recommendations.

Revenue model: Same playbook as Project 1: SEO → Ads → Affiliate → Membership (e.g., paid deep-dive reports on specific models).

3. OrientalMystic — Chinese Metaphysics for Global Audiences

A platform targeting Western users, using AI to generate BaZi (destiny) readings, daily fortune recommendations, and similar content rooted in Chinese metaphysical traditions.

Three reasons this made sense:

  1. Western interest in Chinese culture is growing, and acceptance of mysticism/spirituality is often higher than in China itself;
  2. AI can generate these readings easily—low operational cost, no need to become a "master";
  3. Natural extension into cross-border e-commerce: recommend lucky charms, crystals, or accessories based on readings.

Revenue model: Monthly subscriptions + eventual e-commerce integration.

4. CogNexus — Cognitive Connection Platform

This one is different—fully open source, not primarily profit-driven.

Part One: Cognitive Crystallization / Digital Twin

Beyond just recording facts about you (like ChatGPT's Memory), it analyzes and extracts your cognitive patterns and reasoning habits, forming a digital twin that actually thinks like you and can handle tasks in your style.

Part Two: Digital Twin Collaboration

  • Query celebrity digital twins (imagine asking "Luo Yonghao" a question) affordably, without taking up the real person's time;
  • Hire multiple digital twins to collaborate on predictive tasks—simulate different demographics for sentiment analysis or scenario planning;
  • Even simulate public figures to forecast questions like "Will the Fed cut rates before June 2026?"

The vision: AI that truly thinks like you, delegating more tasks on your behalf. Digital twins collaborating continuously could form something like collective intelligence.

Lessons Learned

1. Token Costs & Budget Management

OpenClaw supports binding fixed quotas (like Claude's subscription tiers), avoiding unpredictable per-call billing. My approach: use part of the quota during focused work sessions, then let OpenClaw run background tasks overnight to use up the remainder. Keeps costs predictable.

2. Project Selection: Just Start

You'll never know the pitfalls until you ship something. Execution cost is actually low—domain and server fees are minimal, and AI tools have dramatically reduced the cost of experimentation.

"Real knowledge comes from doing, not planning."

Pitfalls I hit (Game Strat Port example):

  • I'm not a hardcore gamer, so AI-aggregated guides couldn't truly satisfy dedicated players;
  • Pure SEO growth is painfully slow. Paid acquisition is a clear skill gap I need to address.

3. Choose Projects That Compound

The best projects share this trait: they can run automatically, accumulate value over time, and don't require constant attention.

All four of these qualify:

  • AI monitors model releases → auto-updates LLMPick
  • AI monitors game patches → auto-updates guide content
  • AI generates metaphysics content on schedule → auto-publishes to the platform

The end goal: More projects, less effort per project, more diversified income streams. Assets compound, forming a sustainable long-term trajectory.

Back to Home