Skip to main content

Seriously playing with AI —
showing you how

First-hand AI tool tests and tips
to work smarter and live better

Penchan 🐧

Official OpenClaw 🦞 Contributor · CS Master's student 🎓
A Taiwan-based AI educator — teaching you to go from "just typing prompts at ChatGPT" to "actually getting AI to do the work"

National Tsing Hua University · National Chung Hsing University · National Taipei University of Technology · Soochow University

Meet Penchan →
Penchan scrolling on a phone 🐧
AI LAB

Popular AI Tool Tests & Workflow Guides

Explore AI Lab →
Illustration showing the difference between a fixed workflow and a dynamic workflow handling the same problem (image source: Anthropic) AI Lab 9/23

What Is a Multi-agent Orchestration Framework (Multi-agent harness)? Design Principles and a Real-world Case Study of One Person Directing a Team of AI Agents

A multi-agent orchestration framework (multi-agent harness) lets one orchestrator split work among multiple AI agents, each working in its own context. This plain-English guide covers three typical single-agent failures, six common patterns, when not to use one, five design principles, and how Penchan's AI team used nine AIs to write an article, and what went wrong.
Explore AI Lab →
RECENT PROJECTS

Penchan's Recent AI Projects

View Projects →
AI Agent Memory System (2026) | Complete Guide to Short-Term and Long-Term Memory

AI Agent Memory System (2026) | Complete Guide to Short-Term and Long-Term Memory

AI Lab 5/8

Learn why AI Agents forget previous chats, how short-term and long-term memory differ, and where vector databases, RAG, Dify, and Claude Code fit.
AI Lab 5/8

An AI Assistant That Grows With You: How to Make It More Like You Without Letting It Become Someone Else Next Tuesday

A three-month record of building a personal AI assistant that evolves over time, with failed designs, five surviving abstractions, and the open reference architecture.
AI Lab 4/1

How Should an AI Assistant's Memory Be Designed? Starting from the Claude Code Source Leak

A field note on AI memory architecture after rebuilding an assistant overnight: where to place knowledge so the next run naturally reads it.
AI Lab 3/29

Multi-Agent Collaboration Guide: Stop Your AI Team from Working in Silos

Learn five AI agent orchestration patterns for communication, work division, and failure handling, from file blackboards to circuit breakers.
View Projects →
AI BEGINNER

New to AI?
Start with these 👋

From practical ChatGPT tips to AI presentations and image generation —
curated articles you can apply to your work and life right away 😄

Beginner Guides →
ChatGPT Free Plan Guide (2026) | What Can You Do Without Paying? All Limits Explained

ChatGPT Free Plan Guide (2026) | What Can You Do Without Paying? All Limits Explained

 Gamma vs NotebookLM (now Gemini Notebook): Which AI Slide Tool Fits You?

Gamma vs NotebookLM (now Gemini Notebook): Which AI Slide Tool Fits You?

 AI Image Prompt Techniques 2026 | Chinese Prompts for Precise Gemini and Midjourney Image Generation

AI Image Prompt Techniques 2026 | Chinese Prompts for Precise Gemini and Midjourney Image Generation
Beginner Guides →
AI RESEARCH & ANALYSIS

AI Industry Analysis

Read the AI Industry Reports →
The Anthropic IPO confidential S-1 explained: process, valuation, and access by region

The Anthropic IPO Explained: What a Confidential S-1 Is, What We Know, and How to Participate by Region

Market View 9/21

Anthropic confidentially filed a draft S-1 with the SEC, and OpenAI followed suit. This guide explains the process, confirmed figures, compute expansion, underwriting progress, and how investors in different regions may participate later.
Market View 9/21

Who Invests in Anthropic? Full Investor List, Funding Rounds & Partner Map

Anthropic's investors include Amazon, Google, Salesforce, and NVIDIA—many of whom also supply its compute. A full breakdown of every funding round, key shareholders, and the partnerships that shape the company behind Claude.
Market View 9/21

Anthropic's Valuation Nears a Trillion Dollars: Unpacking Its Funding, Revenue Metrics, and IPO Rumors

Anthropic's valuation jumped from $183 billion to $965 billion within half a year, and there's talk it could go public as early as the second half of 2026. This piece takes an informational approach to unpacking where these numbers come from, how to read ARR, and where the IPO rumors actually stand.
Market View 9/21

How Much Is OpenAI Worth? The IPO Timeline and Making Sense of the "$800 Billion"

OpenAI's valuation nearly tripled in a year, climbing to roughly $852 billion, and in June 2026 the company officially confirmed it had confidentially filed an S-1 for an IPO. This piece uses a research framework to break it down: how the valuation is calculated, where ARR and actual revenue diverge, and why every outlet reports a different IPO timeline.
Read the AI Industry Reports →