Was keen to figure out how AI was actually being used in the workplace by knowledge workers – have personally heard things ranging from "praise be machine god" to "worse than my toddler". So here're the findings!

If there're any questions you think we should explore from a data perspective, feel free to drop them in and we'll get to it!

Notes on the methodology:

  1. We used Search Engine APIs and iterated across search terms related to "AI in the workplace" to find "threads", and removed threads from "overly-specific subreddits" (e.g. coding, cursor, SEO, midjourney etc.) to prevent data skew to get to ~30 subreddits
  2. We then ran those same search queries through SERP API, directly to those same subreddits to find more threads
  3. Scraped comments from those threads, and used a small model to tag it as "relevant" to each question we're trying to answer (we tried the lazy way to just dump all of it into each O3, Gemini Pro, Grok-4, Claude Opus but the checks we placed in indicated that at least some of the results were hallucinated)
  4. For each question, we cleaned the data to reduce context spam and ran that into each model for the analysis (which gave similar answers across the board, and passed our hallucination checks!)
  5. We did light (25 comments) on users too to answer questions like "role" and "is this user just a spammer" and tried to remove those. But some might have gotten past those checks

Posted by yingyn

7 comments

  1. Source: Reddit
    Tools Used: Apify. Graphic by good old powerpoint

  2. Isn’t it possible that there simply are more engineers active on reddit compared to other professions mentioned in the post?

    Can you provide what percent of the users on the subs from which you got your data were engineers/sales/marketing etc?

    And what percent of Redditors are in engineering/sales/ marketing etc. ?

  3. Meeting notes is super useful. It lets me focus on the meeting instead of taking notes 

  4. Good solid clean professional work. Clear labels, easy to understand, discloses data across several visuals to help provide context. Is it artistic and visually stimulating, not really but the design is clean and it is well suited for purpose. Wish this was the norm.

  5. I’m curious how people are going about using it for task management? Is it helping you group or prioritize existing tasks? Giving you daily reminders?

    Also shocked that meeting notes is so low!

  6. 40+% of responses mentioned it’s a pain to deal with hallucinations and quality problems.

    Oh my word. That’s shocking.

    If satnav misdirected you to the wrong place about half the time no one would trust it ever again.

  7. What exactly falls under “long form” and “short form” writing? Internal documents? Emails? Copy writing?

    I’m curious what people are actually using it for!

Comments are closed.