NOW 100% AGENTIC*

SLOPINDEX

Twice a day an agent reads company websites and counts the AI-flavoured filler. Soy Latte Index is dead. Long live the slop.

* the agent is a regex. That is the joke.

01 Podium of shame

02 The full board

Sort by

    03 Buzzword charts

    Total hits across every site scanned. Hover a bar for how many companies use it.

    04 Sloppiest industries

    Average index per category.

    Analog heroes

    The least sloppy sites on the board. Respect.

      05 Buzzword radar

      Twice a day a second agent reads everything the scanner stored and looks for 2- and 3-word phrases that many companies repeat but the index doesn't count yet. Hover a phrase for where it was found.

      06 Refused the bot

      These sites blocked an honest bot that says who it is and reads robots.txt. We respect the no. We also list it.

      07 The Slop Awards

      08 Wall of quotes

      The single sloppiest sentence on each site, verbatim.

      09 Meeting bingo

      Built from the most used buzzwords on the board. Take it to your next all-hands and click whatever you hear.

      10 Scan your own slop

      Paste a LinkedIn post, a job ad or a strategy slide. Runs in your browser only.

      🔖 Slop-o-meterDrag this to your bookmarks bar. On any page, click it to score that page (or just the text you selected). The text goes from your browser to this tab and nowhere else.

      0

      11 Nominate a company

      The board holds 100 companies at most. The agent scans new nominations on its next run (04:30 and 16:30 Helsinki time): up to 25 pages, robots.txt obeyed, careers pages first.

      12 How the index works

      Text

      About 90 buzzword patterns ("harness engineering", "delve", "agentic", "unleash", the "it's not X, it's Y" pivot, em dashes, ✨🚀) as points per 1000 words, times four. Each term counts at most 15 times so a product name can't win alone.

      Signals

      Chief AI Officer, "AI-first" in its own words, a .ai domain, a chatbot widget, job ads asking for agentic or LLM experience, harness-engineering talk.

      Counted

      AI-flavoured job titles on careers pages and products named "… Agent".

      Tiers

      0 Analog hero · 16 Mildly prompted · 41 Slop-curious · 81 Full slop · 151 Slop singularity.

      The agent's shift

      1. 04:30 & 16:30 Wakes up in Helsinki and picks the watchlist plus new nominations.
      2. robots.txt Reads it first. A "no" is a no, and the site is listed as refusing the bot.
      3. Crawl Front page, then about, careers, AI and strategy pages, and job boards linked from careers. Max 25 pages.
      4. Strip Throws away menus, footers, scripts and cookie banners so only the prose counts.
      5. Count Runs 90+ patterns over the text, caps each at 15 hits, checks the signals.
      6. Learn An hour later the radar run looks for new phrases 10+ companies repeat and adds up to 3 per run to the list.
      7. Score Points per 1000 words × 4, plus signals and counts. History keeps the last 30 runs.

      Every buzzword the agent counts

      Points per hit. Sources: lists of overused ChatGPT words, studies of their rise in academic papers, Barron's count of "it's not X, it's Y" in company filings, and LinkedIn.

      Terms

      Signals

        Counted

          Sloppiest pages

            History