Recently, while reviewing the backend data for datafox.tw, I discovered a very harsh reality: some of the hardcore technical articles I spent a great deal of time writing were virtually invisible to search engines and AI crawlers. In 2026, an era where AI SEO and GEO (Generative Engine Optimization) dominate attention, my failure to focus on EEAT was a critical strategic mistake.

🔍 The Fundamental Problem: I’m Trying to Compete with the World’s Strongest Giants for the Same Keywords

I previously wrote an article discussing flash vs thinking kv cache test time compute. I thought this article, being highly technical, would attract decent organic traffic. The result? The data was dead silent.

The reason is simple: this is a purely “topic-based” keyword. When users or AI search for this term, who are the competitors my newly established datafox.tw domain has to face? They are top papers on arXiv, the official technical blogs of Pure Storage, and senior AI researchers on Substack with a large number of accumulated backlinks. In front of these highly authoritative giants, even if my article is well-written, it will simply be drowned out.

In contrast, for searches with my personal tags, such as datafox llm or datafox ntu, I consistently secure the top spot. This reveals a harsh truth: if an article isn’t tied to “who you are,” it’s just a generic popular science article that can be replaced at any time.

🦊 Information is Replaceable, but “Entities” Are Not

In the AI era, the cost of acquiring knowledge has approached zero. If I only write an article on “How to Use GitHub Actions to Automate Website Updates,” AI can easily generate a dozen perfect versions for you.

But what can AI not brute-force learn? It’s contextuality and personal experience.

Taking automated updates as an example, if I write: “How did this fox named Datafox manage to do his weekly reports using this automation system while studying data science at NTU and interning at Google?” This kind of experience, filled with real flesh and blood, cannot be replicated by AI.

In the future AI knowledge graph, I don’t need to be the person who understands GitHub Actions best on the entire internet, but I can precisely claim the unique intersection of “automation + master’s in data science.”

🎤 Monday’s AI Club Class: A Self-Realization Awakening

I profoundly realized this because I happened to lead an AI club class this past Monday.

I spoke on stage for a full two hours, teaching everyone the importance of building a personal brand in 2026 and how to use AI Agents for marketing. But when I stepped off the stage and re-examined my own website, I realized with a wry smile: although I spoke eloquently on stage, my recently produced technical articles closely resembled soulless AI-generated content.

I taught everyone to become “mountain-shaped talents” in this era—meaning to treat skills that AI can easily boost to a 60% level (like writing basic code, generating marketing copy) as a broad base, and then to refine two to three unique experiences that only you possess, converging into a “mountain peak” that others cannot brute-force with AI.

In my case, this mountain peak should be the chemical reaction of “AI Technical Prowess × Financial Foundation × Public Speaking and Team Leadership.”

Yet, I forgot to showcase this “mountain peak” of my own on my digital territory. I wrote a lot of pure technical analyses that AI could also generate, trying to compete with global giants for SEO, but I completely obscured the most important aspects: “who I am” and “why I encountered this problem.”

(Regarding the concept of “mountain-shaped talents” and how to build a career moat in the AI era that won’t be eroded by inflation, there are many hard-won insights to share. I will write a dedicated article later to delve deeply into this, so please look forward to it.)

🛠️ Future Correction Strategy: Weaving “Me” into Every Technical Article

With painful reflection, and to reclaim the soul of my articles and the weight of E-E-A-T, I have decided to fully refactor my future technical writing strategy:

1. Short-term: Abandon Topic-Based Titles, Bind to Unique Tags

I will no longer use generic titles like Flash vs Thinking 模式的 KV Cache 實測 (Flash vs Thinking Mode KV Cache Practical Test). Future titles must inherently carry my Entity perspective. For example:

  • “From a GenAI Intern’s Perspective: Flash vs. Thinking Mode KV Cache Practical Test”
  • “Test-Time Compute Trade-offs I Encountered in the Google Pixels Team”

This way, when someone searches for “GenAI intern KV cache,” the sole and most authoritative answer will be me.

2. Mid-term: Let Technology Become an Extension of Personal Tags

The end of every technical article must anchor back to my real experiences. It will no longer be just a cold conclusion; for example, I need to add:

  • “This architectural problem is precisely where students most often get stuck when I teach at the NTU AI Club…”*
  • “This practical test is actually the most painful pitfall I personally encountered when deploying LLMOps at Cathay United Bank…”*

This will allow AI, when indexing, to firmly bind these hardcore technical aspects to the entity “柯宥圻”.

3. Long-term: Revitalize Medium’s Historical Assets

My old articles written on Medium have actually accumulated decent traffic and Domain Authority. I will start adding backlinks to datafox.tw within those articles, gradually transferring the accumulated authority to this digital sovereign territory that truly belongs to me.

In summary: No matter how good the content of a technical article is, if it lacks a soul, it will merely be an “anonymous article about a certain technology” in the eyes of AI and search engines. Starting today, I will unreservedly weave my hard-won experiences and struggles into every technical article. After all, this is the true essence of a fox constantly debugging its life in beta.