About

I spent a decade in B2B marketing, helping companies figure out who their customers are and how to reach them. Then I watched AI change the game—and decided to build instead of just analyze.

The GTM years

My career has been about understanding markets and positioning products. I've built GTM strategies for B2B SaaS companies, figured out how to explain complex products to the right people, and learned that most marketing problems are actually product problems in disguise.

Along the way, I became one of Vietnam's top SEO practitioners for English B2B markets. Not because I love algorithms, but because understanding search is understanding intent—and intent is the purest signal of what people actually want.

The pivot

When ChatGPT launched, I saw something shift. Marketing intelligence—my entire domain—was being reinvented around LLMs. I could keep being a "AI-powered marketer" or I could learn to build the actual tools.

So I started coding. Linux, Python, RAG pipelines, fine-tuning. The learning curve was brutal, but the leverage was obvious. I could now prototype ideas in an afternoon that used to require a team and a budget.

What I'm doing now

I'm building at the intersection of GTM and AI. Specifically:

  • GTM intelligence systems — Using agents to analyze competitive positioning, messaging gaps, and market opportunities at scale
  • Agentic workflows — Exploring how AI agents can collaborate in knowledge work environments like Obsidian
  • Local AI infrastructure — Figuring out what actually works when you run everything locally (models, tooling, tradeoffs)

The philosophy

I believe the best tools are built by people who actually use them. That's why I write about my experiments in public—failures included. The AI landscape changes weekly, and the only way to keep up is to ship things, break them, and learn.

Building this site

This website is itself an experiment in AI-assisted development. I designed the architecture and wrote the spec, but Claude did most of the actual coding—component scaffolding, Tailwind configuration, Astro setup.

The goal wasn't to outsource the work. It was to model what thoughtful AI collaboration looks like: I provide direction, taste, and constraints; the AI handles implementation speed. We iterate together. That's the future I'm building toward.

Want to collaborate or just chat? Email me or find me on Twitter.