Field log · Writing
92 essays, 90,512 words
I write about generative AI in Traditional Chinese first, for readers in Taiwan. These are the English translations. Pick a region to see what I’ve written there.
2026.09.23 Memory Summit Jev Lab: Deciding What an Agent Should Remember 436 words 2026.09.22 Memory Summit Remembering Isn't the Same as Remembering Right 3,936 words 2026.09.04 Memory Summit How Agent Memory Forms, Updates and Fades 3,230 words 2026.08.24 Memory Summit Everyone Talks About Memory. Do We Mean the Same Thing? 1,881 words 2026.08.24 Memory Summit What Does an Agent's Memory Actually Look Like? 2,601 words 2026.08.24 Memory Summit What Should an Agent Remember? 3,027 words 2026.08.02 MCP Pass MCP Goes Stateless: Getting Ready for Scale 1,947 words 2026.07.21 Memory Summit AWS Summit × Agile Taipei: Two Packed Talks on Agent Memory 309 words 2026.07.05 MCP Pass Productizing MCP: Where to Draw the Line as Tools Pile Up 2,311 words 2026.06.09 Agent Range Employee, Assistant or Tool? How Three Agent Frameworks Really Differ 1,162 words 2026.05.29 Agent Range Speaking at WiDS 2026: Five AIs on My Team 504 words 2026.05.22 Memory Summit How We Classify Knowledge, and How That Keeps Changing 1,936 words 2026.05.10 Agent Range AI Agent Security Is Less Mysterious Than You Think 1,363 words 2026.04.28 Agent Range Anthropic's Global Hackathon: Building Cairn Solo in Six Days 932 words 2026.04.18 OpenClaw Bay Interviewed by Business Weekly: Why OpenClaw 1,323 words 2026.04.17 Agent Range Lessons from Building Multi-Agent Systems 720 words 2026.04.14 Agent Range Harness Engineering: Notes on Hung-yi Lee's New Course 666 words 2026.04.12 LLM Plateau Do AIs Have Emotions? Reading Anthropic's New Research 492 words 2026.04.06 Agent Range agent-broker, Tested: Letting Agents Talk on Discord 244 words 2026.04.05 Agent Range Meeting Pahud: AI Agent Trends and the Business of MCP 479 words 2026.04.05 OpenClaw Bay On USTV News: OpenClaw Use Cases, Risks and NemoClaw 354 words 2026.03.26 Agent Range The Agent Economy Is Taking Shape 584 words 2026.03.26 OpenClaw Bay Running OpenClaw Without Paying: The API Sources I Use 1,117 words 2026.03.26 OpenClaw Bay Raising Your OpenClaw Agent: The Tools 592 words 2026.03.20 OpenClaw Bay How I Design Skills and Workflows on OpenClaw 711 words 2026.03.15 OpenClaw Bay My First TV Interview and Studio Recording at PTS 383 words 2026.03.04 OpenClaw Bay A New Way to Use Antigravity for Free 314 words 2026.03.02 Memory Summit AI Doesn't Lack a Brain. It Lacks Memory That Builds Up 542 words 2026.02.10 Vibe Coding Valley At R-Ladies Taipei: From Side Project to Business 302 words 2026.02.05 OpenClaw Bay Four Pitfalls of Connecting OpenClaw to LINE 1,021 words 2026.01.28 Agent Range StepFun Open-Sources Step-DeepResearch 158 words 2026.01.20 Vibe Coding Valley Don't Fall into the Anti-AI Hype 789 words 2026.01.19 Vibe Coding Valley Minimal AI Coding: First Principles Over Frameworks 1,704 words 2026.01.18 Agent Range The Week AI Agents Started to Converge 556 words 2026.01.16 OpenClaw Bay On Cowork and Skills 682 words 2026.01.15 OpenClaw Bay The Networking Pitfalls Beneath OpenClaw 424 words 2026.01.11 Agent Range AI Agents Are Becoming Infrastructure 731 words 2026.01.08 LLM Plateau Notes on Karpathy's 2025 Review 47 words 2026.01.08 OpenClaw Bay One Week with OpenClaw: Getting Started 248 words 2026.01.05 LLM Plateau 2025 in Review: LLMs Started Doing Real Work 241 words 2025.12.28 Vibe Coding Valley Rethinking Prompt Caching Through ngrok's Article 187 words 2025.12.20 Vibe Coding Valley Professional Software Developers Don't Vibe, They Control 51 words 2025.12.18 Vibe Coding Valley After Observability, the Next Step Is Evals 979 words 2025.12.15 LLM Plateau Pretraining from Scratch to Production: What I Learned 1,018 words 2025.12.14 Agent Range I'm Starting to Believe AI Agents Will Do Real Work 690 words 2025.12.12 Agent Range State of Agent Engineering 2025 by LangChain 92 words 2025.12.09 Vibe Coding Valley A Real Botnet Encounter: The React Vulnerability Rated CVSS 10 674 words 2025.12.08 Vibe Coding Valley AI Observability Is Becoming Standard 415 words 2025.12.05 Vibe Coding Valley How Google Came Back 354 words 2025.12.01 LLM Plateau The Multi-TW Dataset 802 words 2025.11.20 LLM Plateau Prompt Repetition Improves Non-Reasoning LLMs 199 words 2025.11.15 Agent Range Bringing AI into Real Industries 465 words 2025.11.10 Vibe Coding Valley Why uv Is So Fast 49 words 2025.11.05 Vibe Coding Valley Do You Review All the Code AI Writes? 326 words 2025.10.20 Voice & Avatar Hills The Rise of Voice AI: Machines That Listen and Talk Back 1,935 words 2025.10.20 Voice & Avatar Hills D2E: Training Embodied AI on Desktop Data 189 words 2025.10.08 LLM Plateau Answer, Refuse or Guess? Teaching Language Models About Risk 1,163 words 2025.10.05 LLM Plateau Answer, Refuse, or Guess? Reading CK Wu's Paper on Risk-Aware Decisions 791 words 2025.09.26 Vibe Coding Valley AILogora Is Live: Our Stack and the Pitfalls 194 words 2025.08.28 Voice & Avatar Hills VibeVoice: Microsoft's Open-Source TTS 191 words 2025.08.22 Vibe Coding Valley Architecture Traps in the Age of Vibe Coding 357 words 2025.08.15 Vibe Coding Valley Vibe Coding Tools 225 words 2025.06.06 Vibe Coding Valley Migrating Away from v0: Switching to Cursor and Windsurf 686 words 2025.05.14 Agent Range How Do AIs Talk to Each Other? A Look at Agent Protocols 2,709 words 2025.03.15 Agent Range OpenManus Tutorial: How to Build Your Own AI Agent 1,105 words 2025.03.05 LLM Plateau Understanding Reasoning Models and Test-Time Compute: Insights from DeepSeek-R1 4,192 words 2025.02.16 LLM Plateau Running DeepSeek Locally with llama.cpp and Ollama 1,484 words 2025.02.03 Voice & Avatar Hills AI Avatars: How Virtual Characters Are Made 2,004 words 2024.11.06 RAG Basin Local GraphRAG with llama.cpp 880 words 2024.07.19 RAG Basin Knowledge Graph + RAG: Microsoft GraphRAG, Implemented and Visualized 1,370 words 2024.06.17 Voice & Avatar Hills Building an Interactive AI Assistant: LLMAvatarTalk 257 words 2024.06.05 RAG Basin RAG Evaluation Frameworks Compared: TruLens, RGAR and RAGAs 2,120 words 2024.04.12 LLM Plateau Evaluating LLMs with EleutherAI's LM Evaluation Harness 1,081 words 2024.03.29 LLM Plateau A Guide to LLM Evaluation: Trends, Metrics and What's Next 1,217 words 2024.03.18 LLM Plateau Comparing LLM Specs: ChatGPT, Gemini, Claude, Mistral, Llama and More 580 words 2024.03.10 LLM Plateau A Guide to Prompt Engineering Techniques 4,147 words 2024.03.07 RAG Basin Seven RAG Challenges and How to Solve Them 1,979 words 2024.01.27 LLM Plateau Short on Memory? LLM Compression Techniques 1,942 words 2024.01.25 RAG Basin RAG Tutorial: Streamlit + LangChain + Llama 2 477 words 2024.01.20 RAG Basin RAG Tutorial: Build Your Own LLM with LangChain and Llama 2 1,026 words 2024.01.16 RAG Basin RAG: Retrieval-Augmented Generation Opens a New Chapter for NLP 2,448 words 2024.01.15 LLM Plateau Run Llama 2 on Your Phone: A llama.cpp Tutorial 808 words 2024.01.15 LLM Plateau How to Download Models from Hugging Face 425 words 2024.01.14 LLM Plateau Build Your Own LLM: A Llama 2 Tutorial 381 words 2024.01.13 LLM Plateau Why Pay? GPT-3.5 vs GPT-4 604 words 2024.01.12 Data Lowlands Your First Open-Source Pull Request: A Beginner's Guide 652 words 2024.01.10 Data Lowlands Six Clustering Algorithms You Should Know 1,408 words 2024.01.08 Data Lowlands Six API Architecture Styles You Should Know 556 words 2023.12.30 Data Lowlands Faster Pandas: Installing and Using Polars 409 words 2023.12.30 Data Lowlands Faster Pandas: Installing Modin 361 words 2023.12.29 Data Lowlands Faster Pandas: Installing cuDF on WSL2 334 words 2023.12.28 Data Lowlands Installing conda on WSL2 525 words
Nothing in this region yet.