Henry Nguyen

Founder, Xiren

Posts

• Your CRM Is Not Broken. Your Workflow Is.

  Most CRMs do not fail because the software is bad. They fail because updating the CRM is treated as a second job after the real work. If the system of record is not built into intake, routing, handoffs, and follow-up, the data will rot.

• The Automation Gap

  I wanted to test whether AI could break a golden rule of operations — that you can't scale a mess. So I built a simulation: same orders, same model, four versions of the same delivery company, varying only the cleanliness of inputs and whether a human or AI was making the calls. The condition where AI met chaos performed worse than chaos alone.

• An AI-Enabled VA Team Will Outcompete a Mid-Market US Team. The Only Question Is When.

  What an AI-enabled VA can do in 2026 is meaningfully different from what a non-AI-enabled VA could do in 2022. The capability gap that protected mid-market US knowledge workers is the layer AI is best at scaffolding. The economics will not care about how anyone feels about it.

• Build Log: A Google Business Lead Tool, And The Sneaker Bot Brain That Almost Built It Wrong

  Spent the first hour researching residential proxies, stealth plugins, and request jitter for a tool that needed none of it. The legitimate path was a Google API and a credit card. Adversarial mode is occasionally the right answer. It's almost never the first one.

• Where AI Lands First Won't Be the Rich World

  The global learning curve on what works with AI is going to be flatter than people expect. The most interesting deployment case studies of the next decade won't come from American Fortune 500s.

• Tech Moves Exponentially. Operations Move Linearly. The Gap Is Where Value Lives.

  Frontier models double in useful capability every 6 to 12 months. Operations don't move at that speed and never will. The gap is where the next decade of value lives.

• Local Models Just Caught Up. The Industries That Couldn't Use AI Yesterday Can Use It Today.

  Regulated operators kept stalling at the same point: the data can't leave the building. With Gemma 4, that excuse is gone. The model can live inside your network and run against your data without any of it leaving.

• Why AI Implementations Fail

  The technology works. The implementations fail. Both are true. Here's the structural failure pattern, why operators keep walking into it, and the four-question diagnostic that separates workflows that are ready for automation from ones that aren't.