Yesterday on X, Aadit Sheth posted that Perplexity quietly released a 42-page internal guide on how they use AI in their work. The post included several screenshots of the guide, which looked pretty legitimate.


He mentioned the most useful points were: using AI to automate chores (emails, meeting prep, research), using AI to amplify curiosity rather than replace it, and their simple, practical prompt handbook. The post ended by asking people to comment "AI" to get the guide via DM.

But soon after, the plot took a turn.

First, someone posted Perplexity AI's own search results, showing that as of October 2025, the official site had not released a guide called "Perplexity at Work." The so-called guide was actually educational content from an independent publication.


Next, people in the comments started directly sharing the PDF link: https://r2cdn.perplexity.ai/pdf/pplx-at-work.pdf. Many pointed out that this was never a confidential internal document but just an open resource on Perplexity's official website, part of product promotional material.

Some called it "engagement bait"—using DMs to send out public links to elicit comments and interactions. "Why would you try to control information about something you didn't create?" one comment asked.

The story is simple: an open product guide was packaged as "internal material" to grab attention. Such tactics are far too common on social platforms.

But setting aside the marketing angle, does the guide itself have value? I downloaded it to take a look.


Inside the guide, there are indeed some concrete prompt examples, like how to have AI summarize the past four hours of emails and highlight urgent matters, how to query industry key news, and how to fetch status updates from tools like Salesforce and Jira. For those unfamiliar with AI tools, these are decent introductory references.

But calling it a revolutionary internal secret recipe would be an overstatement. It's more like a product manual, with the core purpose being to showcase Perplexity's features and guide users toward their enterprise service.

What's most interesting about the whole affair isn't the guide's content but the small-scale online interaction itself—the rapid verification of information, the ability to see through common marketing gimmicks, and the skill to maintain basic judgment in an information overload environment.

As for whether AI makes us smarter or lazier, it may depend on how we use it. Tools themselves are neither good nor bad, but how we use them often reflects our intentions.