Why I build the tools I need

There’s a thing that happens in product. Sales or AM comes to you and says: everybody is asking for this feature. You ask who. They give you a look. You build the feature anyway because the signal feels real even if the documentation doesn’t exist.

Then you ship it and want to run demos. You go back to sales and AM: hey, who were those customers who wanted this?

They don’t remember.

This is not a complaint about sales and AM teams. This is just how it goes.

The thing is, we record every sales call. Every single one, in Jiminny. The customers who asked for this feature are in there somewhere — their names, their exact words, what problem they were trying to solve. The data exists. The search just doesn’t work. You type in a keyword and get back something that might generously be described as results.

I needed those calls. I had Claude Code and a coworker in engineering willing to give me access.

So I built Wisp — a search layer over our Jiminny exports that actually works. Indexes the transcripts, filters by keyword, rep, date range, deal stage. You put in what you’re looking for and it finds it.

It took a long flight over spring break. The data was already there. The problem was never the data.

I think about this a lot when people ask about AI tools and whether PMs should be able to code. The question I actually ask is: when something is broken and you have the means to fix it, what do you do? I had Claude Code, I had access, I had a long flight. Waiting for a ticket to get prioritized wasn’t going to find me those customers.

Wisp isn’t a product. It’s a tool I built because I needed it. That’s usually the best reason to build anything.