How We Rank Bots

What The Score Uses

• A clear job (45 points). How clearly the bot's own description says what job it does and what you get back, and how much detail its maker wrote.
• A complete page (20 points). How many "What It Does" parts we could fill from the bot's own words: who it's for, what you ask it, what you get back, what it connects to, and what it won't do.
• Asks before acting (20 points). Full marks when it drafts or asks before it sends, posts, buys or deletes. Low marks when it says it acts on its own. A middle score when it does not say.
• Likes on X (15 points). Only for Swarm School library bots, where we track likes on the bot's post. Bots made by Grok Bot users have no like data, so their other three parts are scaled up to 100. Likes can only help: a library bot gets the better of its score with likes and its score without them.

Look-Alike Bots

Bots with nearly the same description are grouped. The best one leads and shows "+N similar". The others lose 8 points and sit behind it.

Where You See It

• Order: a job group's Superbots come first, since each one does the whole job. Then every bot, best first. Library bots and bots made by Grok Bot users are ranked together.
• Top Pick: the top 3 bots in each job group's order that are not look-alikes.
• Top 10: the first page of each group lists its 10 best bots.
• Search: the best match comes first. Ties go to the higher score.
• Bot Universe: each job group is a galaxy. Library bots form its core, best first; bots made by Grok Bot users ring around it, best closest. Higher scores get a bigger, glowing bot.

What It Does Not Mean

We score only what each maker wrote. x.ai shows a bot's name, maker and description, not its instructions or connectors. A high score means a clear, careful description. It does not mean Swarm School or xAI made, tested or endorses the bot. Read a bot's page before you add it.

Browse the bot library · Productivity & Work, best first