---
title: "Comment by Hafiz Waleed Ahmed on \"Do AI assistants recommend your app? I built a small weekly tracker to find out\" | Hashnode"
description: "Separating named from cited, and keeping the by-name questions out of the headline number, is what makes the 4 of 72 figure readable. A trend only works if the"
url: "https://hashnode.com/posts/do-ai-assistants-recommend-your-app-i-built-a-small-weekly-tracker-to-find-out/6b00ffad8ef3e6ec53a2746b/comment/6b01539926b829fc64b359de"
---

[Discussion](https://hashnode.com/posts/do-ai-assistants-recommend-your-app-i-built-a-small-weekly-tracker-to-find-out/6b00ffad8ef3e6ec53a2746b)

Comment [Do AI assistants recommend your app? I built a small weekly tracker to find out](https://hashnode.com/posts/do-ai-assistants-recommend-your-app-i-built-a-small-weekly-tracker-to-find-out/6b00ffad8ef3e6ec53a2746b)

[Hafiz Waleed Ahmed](https://hashnode.com/@hafizwaleedahmed5)

[1d ago](https://hashnode.com/posts/do-ai-assistants-recommend-your-app-i-built-a-small-weekly-tracker-to-find-out/6b00ffad8ef3e6ec53a2746b/comment/6b01539926b829fc64b359de)

Separating named from cited, and keeping the by-name questions out of the headline number, is what makes the 4 of 72 figure readable. A trend only works if the original questions stay fixed, which you called out, because a reword hides whether the assistant changed or the prompt did. The cost jump when Claude search results come back as input tokens is a useful warning if someone starts from a cheap first estimate.
