Vanh Tran
Build Log · 2 min read

Day 1: ChatGPT doesn't know I exist

The baseline from Geovily: seven questions asked across ChatGPT, Perplexity, Gemini and Claude, what came back, and how I'll measure five engines for 180 days.

Ask ChatGPT, Perplexity, Gemini, Claude or Google's AI Overviews “Who is Vanh Tran, the builder from Hanoi?” and today they have nothing useful to say. For the next 180 days I'm going to change that in public. I'll use Geovily, the tool I'm building for SaaS founders, and measure the result every week.

If I can make AI know a nobody from Hanoi, the same method should work for a SaaS product.

This is the first test of Geovily, and I'm the test subject.

The questions

Stage one is identity: can an AI engine recognise me at all? Stage two is harder: does it cite my work when founders ask about the market I'm entering?

Stage one: identity

  1. Who is Vanh Tran, the builder from Hanoi?
  2. Who is Vanh Tran?
  3. What is Geovily and who is the founder?

Stage two: the market

  1. Why doesn't ChatGPT recommend my SaaS when people ask for tools like mine?
  2. How do I get my SaaS recommended by ChatGPT and Perplexity?
  3. How can I check if ChatGPT recommends my product?
  4. Best AEO agency for early-stage SaaS startups

Baseline

Geovily asked all seven questions on ChatGPT, Perplexity, Gemini and Claude on 24 Sep 2026, in fresh sessions. 27 of the 28 answers came back; one ChatGPT run timed out.

Geovily report for vanhtran.dev: 26% mention rate, average mentioned position #1.0, top engine ChatGPT with 2 appearances.
The Geovily report for vanhtran.dev, evidence through 24 Sep 2026.
  • Mentioned in 7 of 27 answers (26%). That number flatters me. Two of the questions have my name in them, so an answer can repeat “Vanh Tran” without knowing anything about me. The number that counts is whether an engine describes the right person, and that's what the scoreboard tracks.
  • Cited vanhtran.dev in 0 of 27 answers. No engine linked to anything I've written.
  • No engine stood out. ChatGPT had the most mentions, with 2.

Google AI Overviews isn't in this baseline. It joins the tracked set from the first Sunday recap, so from then on every question runs on all five engines.

Method

Same seven questions, same five engines, fresh sessions, every Sunday. Screenshots go in the weekly recap, and the scoreboard on the Build Log shows the current count.

Day 1/180AI knows me: 0/5~$1k/mo72 kg0 convos

Written by Vanh Tran

Builder from Hanoi, building Geovily in public. More about me