I built an AI visibility tracker, then stopped using it myself
I've been building on the internet for almost 20 years. Product, engineering, AI: I've done each of them hands-on, one step at a time. I've shipped SaaS for a traditional industry, for the pet industry, for generating game assets, and for game marketing materials.
The full-stack edge I was proud of (product design, online marketing, engineering) is slowly being diluted by AI. Anyone can ship now. How a product gets sold matters more than it ever did, and selling is my weak spot.
It showed worst with the traditional-industry product. I could barely get in front of the people who'd use it, and some days I honestly couldn't tell whether I was building the right thing.
So I started holding every idea to three rules:
The core of it has to be something I already know deeply and am good at.
I have to be one of its customers. With the traditional-industry product and the asset generators, I never was.
What it delivers needs a ruler. Good or bad can't depend on who's looking. With a ruler, I can point everything I know at moving the product toward "good".
Before Passcite I shipped a first version: an AEO tool that reported whether AI assistants mention your site, and kept tracking it.
Rule 2 is what caught it. I was its customer, and I didn't want to open it. It told me my AI visibility was bad. I knew that. What I needed was how to make it better, and it had nothing to say. A number you can't move is a number you stop looking at.
So I rebuilt the whole flow as Passcite:
Analysis: the buyer questions you lose across ChatGPT, Perplexity and Gemini, and who shows up instead on each one.
Why: what those answers cite, and what your own site never says.
How: every change, split in two. What we can draft for you, and what only you can do.
Action: you approve or reject each draft, tick off the rest, then re-run the same questions.
The split mattered more than I expected. A lot of what the assistants read isn't on your website at all. For every dental practice I've run it on, the answers cite Zocdoc and Healthgrades. A tool whose only output is a blog post has nowhere to put "claim your Healthgrades profile", so the biggest item falls off the list.
Rule 3 applies to me too, so here's my own reading. On September 21 I ran passcite.com through 30 buyer questions. 21 of them don't contain the word "Passcite", and on those 21, none of the three assistants mentioned Passcite. Not once. Every run since mid-September says the same. It did come up on the other 9, but every one of those was a question with my name already in it, like "alternatives to Passcite". That's being told, not being found.
What shows up instead: Semrush on 16 of the 30 questions, rankability.com on 20. And the biggest item on my own list isn't a page. It's a Capterra profile: the assistants cite capterra.com on 16 of the 30. Nobody can do that one for me, my own tool included.
That's the first change I'm making. When it's done I'll re-run the same 30 questions and post the result here, whether or not anything moved. My whole product rests on one thing: that changes like this make an assistant change its answer. I haven't proven that yet. One before-and-after won't prove it either, because the assistants drift on their own, but it's more than I have today.
Something I didn't expect: rule 3 is also doing some of the selling I'm bad at. When I email an agency now, there's no pitch. I run one of their clients' sites and send the numbers: 30 questions a patient would type, 22 of them never say the practice's name, and it came up in 2 of those 22. The number does the talking, or it doesn't. I've sent two so far. Ask me in a week.
Rule 2 only half holds this time, though. I'm a customer as a site owner. I'm not an agency, and agencies are who I most want to sell to. That half I'm learning by asking.
So, one question for each side:
If you run an agency: have your clients started asking about ChatGPT or Perplexity yet, or is it still not on their radar?
If you've built something and then stopped using it yourself: what was missing?
The free report needs no signup: https://passcite.com/free-report
Interesting take. Would you still recommend this approach to someone starting today?
Solid lesson. Which channel has worked best for you so far?
Thanks for sharing the numbers, that makes it much easier to follow.
Interesting approach. What was the hardest part to get right?
Interesting. How are you measuring whether it is working?
Really relatable. How much time do you put into this each week?
Good point. Did you test that with users before committing to it?
How did you decide this was worth building in the first place?
Nice progress. What is the next thing you are focusing on?
Nice progress. What is the next thing you are focusing on?
Interesting. How are you measuring whether it is working?
Appreciate the honesty here, most people only share the wins.
Good point. Did you test that with users before committing to it?
Interesting approach. What was the hardest part to get right?
Interesting approach. What was the hardest part to get right?
Good point. Did you test that with users before committing to it?
What made you pick this stack over the alternatives?
Interesting approach. What was the hardest part to get right?
Appreciate the honesty here, most people only share the wins.
What made you pick this stack over the alternatives?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Interesting. How are you measuring whether it is working?
Interesting approach. What was the hardest part to get right?
Interesting. How are you measuring whether it is working?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Interesting. How are you measuring whether it is working?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Good write-up. What would you do differently if you started again?
Appreciate the honesty here, most people only share the wins.
Clear and practical, thanks. Did anything surprise you along the way?
Good point. Did you test that with users before committing to it?
What made you pick this stack over the alternatives?
Interesting. How are you measuring whether it is working?
Good point. Did you test that with users before committing to it?
With two agency outreaches underway, what response would distinguish genuine recurring client demand for AI visibility from curiosity about a new reporting metric?
Interesting. How are you measuring whether it is working?
Interesting approach. What was the hardest part to get right?
Good point. Did you test that with users before committing to it?
Appreciate the honesty here, most people only share the wins.
Good point. Did you test that with users before committing to it?
Good write-up. What would you do differently if you started again?
Clear and practical, thanks. Did anything surprise you along the way?
What made you pick this stack over the alternatives?
Interesting approach. What was the hardest part to get right?
Interesting approach. What was the hardest part to get right?
Appreciate the honesty here, most people only share the wins.
Good write-up. What would you do differently if you started again?
What made you pick this stack over the alternatives?
Clear and practical, thanks. Did anything surprise you along the way?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Interesting approach. What was the hardest part to get right?
Interesting. How are you measuring whether it is working?
Love this angle. Building Xstream4K right now so this hits close to home — what made you look into it in the first place?
This is great work — reminds me of some of the calls I've had to make building Xstream4K. What would you do differently if you started over?
Appreciate the honesty here, most people only share the wins.
Appreciate the honesty here, most people only share the wins.
Dogfooding is the real filter. The split between buyer questions and what assistants actually surface feels especially useful. What made Passcite worth keeping in your own workflow when the first version failed the open it test?
The split that mattered is a different one: what a tool can draft for me, and what only I can do.
The first version stopped at a number. This one ends in a list I can work through and check again
Interesting approach. What was the hardest part to get right?
What made you pick this stack over the alternatives?
Good write-up. What would you do differently if you started again?
Interesting. How are you measuring whether it is working?
Clear and practical, thanks. Did anything surprise you along the way?
Good point. Did you test that with users before committing to it?
Appreciate the honesty here, most people only share the wins.
Interesting approach. What was the hardest part to get right?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Good point. Did you test that with users before committing to it?
Interesting. How are you measuring whether it is working?
Good write-up. What would you do differently if you started again?
Interesting approach. What was the hardest part to get right?
Clear and practical, thanks. Did anything surprise you along the way?
What made you pick this stack over the alternatives?
Interesting approach. What was the hardest part to get right?
Appreciate the honesty here, most people only share the wins.
Really good writeup, thanks for sharing it. What's the next thing you're planning to try here?