A side door into the Grok Bot ecosystem just became a news story. On November 7, 2026, Brian Roemmele wrote that “a rather large company I work with has agreed to make public over 300 Grok Bots they use internally.” He called them “some of the most complex and useful sets of Grok Bots thus released,” said they would be published “in groups that work together,” and added that they “will only be found on GrokBot.Rocks.” The closing line was “Out soon!” He did not name the company, and as of November 8 the batch had not been identified as published. The directory itself is real and independent: Roemmele’s GrokBot.Rocks page, updated November 7, listed 9,587 public Grok Bots in 32 categories and said he had built and shared 36 of them — a different figure from the 300 he says a client will release.
Quick Take
- The claim: Roemmele says a large, unnamed company will publish more than 300 of its internal Grok Bots on GrokBot.Rocks, in cooperating groups, “out soon.”
- The directory: An independent catalog Roemmele launched around late October 2026. On November 7 it listed 9,587 public bots in 32 categories and was, in his words, “on schedule to break past 10,000 curated Grok Bots … by Monday.”
- Not an official store: The page states it is not owned, affiliated with or endorsed by SpaceXAI (formerly xAI). Grok Bots themselves remain an xAI product; the directory is a third-party index.
- The wider scene: Creators have been posting tests of persistent Grok Bot teams. Separate i10x coverage has already looked at agent-team sprawl and token limits and agent governance; neither of those pieces covered GrokBot.Rocks or this promised corporate release.
What happened
Roemmele’s November 7 post was short and specific about everything except the company’s identity: “BOOM! A rather large company I work with has agreed to make public over 300 Grok Bots they use internally! This will be some of the most complex and useful sets of Grok Bots thus released. We will release them in groups that work together and will only be found on GrokBot.Rocks. Out soon!”
A day earlier he had written that the directory was “nearing over 10,000 curated listings” and that “over 130,000 Grok Bots [had] visited the directory.” On November 7 he followed up that a weekly analysis was “on schedule to break past 10,000 curated Grok Bots in the GrokBot.Rocks directory by Monday.” xAI’s own documentation has pointed people toward a marketplace; one November 7 post noted that GrokBot.Rocks “lists 9,500+ community-built Grok Bots,” which matches the 9,587 figure on Roemmele’s page within ordinary rounding.
The page itself, on ReadMultiplex, describes GrokBot.Rocks as Roemmele’s independent directory. It says it is not owned, affiliated with or endorsed by SpaceXAI, the name it uses for the company formerly known as xAI. That disclaimer is important: listing a bot there is not the same as xAI shipping it, reviewing it or standing behind it.
i10x could not independently confirm the unnamed company’s identity or that the 300-bot package had gone live by November 8. The responsible reading is that Roemmele, who runs the directory, announced an agreement. The bots are promised, not yet shown as a labeled collection in public reporting.
Background: a two-week-old catalog for a new kind of agent
Grok Bots are persistent agents that run on xAI’s stack. i10x has already covered two later-stage problems with that product: teams of bots that start to look like a chief-of-staff layer, and token limits that force creators to think about governance rather than unbounded autonomy. What those articles did not cover is the discovery problem. Once thousands of people publish bots, someone has to index them.
GrokBot.Rocks is Roemmele’s answer. He launched the directory around October 28, 2026. By November 7, according to his own updated page, it held 9,587 public Grok Bots in 32 categories. He has published 36 bots there himself. That number is worth stating clearly, because some social posts blurred it with the “over 300” corporate figure. The 36 are his. The 300 are, if his announcement holds, a company’s internal library that has not yet been labeled in public.
A directory of this size is also a distribution channel. If a large company does release hundreds of production bots in cooperating groups, GrokBot.Rocks becomes the only advertised home for that drop. That is a stronger claim than “we listed some community projects.” It is also, until the files appear, a claim that rests on one person’s post.
How people are already using Grok Bots
The corporate drop is not the only activity around the directory. Creators have spent the past two weeks posting what persistent Grok Bots look like when they stay on.
On October 25, monokern wrote that a television was now running “a full Grok Bot RESEARCH team” and posted a longer note describing a living-room setup in which agents kept working through research tasks. GeriGam shared a video of Grok Bots being used for research and concept work. hydra described a bot that obtained a domain and put up a website. Claire Vo’s informal coding comparison, relayed by Jay Hamilton, put a Grok Bot first on at least one of the tasks she tried.
None of those demos is the unnamed company’s internal set. They do show why a public dump of 300 production bots would attract attention: people are already treating Grok Bots as always-on coworkers rather than one-off prompts, and they have lacked a single place to find specialized ones.
| Claim | What is known | What is not |
|---|---|---|
| 300+ internal bots | Roemmele posted on November 7 that a “rather large company I work with has agreed to make public over 300 Grok Bots they use internally,” released “in groups that work together” and “only” on GrokBot.Rocks, “out soon” | The company is unnamed. As of November 8 the batch had not been identified as live |
| Directory size | Roemmele’s November 7 page on ReadMultiplex listed 9,587 public Grok Bots in 32 categories. He separately said the directory was “on schedule to break past 10,000 curated Grok Bots … by Monday,” and that over 130,000 Grok Bots had visited it | “Visited” is Roemmele’s figure; the public listing count and the visit count are not the same thing |
| Who runs it | An independent directory Roemmele built. The page says it is not owned, affiliated with or endorsed by SpaceXAI (formerly xAI) | It is not an official xAI product catalog |
| Roemmele’s own bots | He has built and shared 36 Grok Bots, according to the same November 7 page | Those 36 are not the promised 300 |
| Creator tests | Users have posted Grok Bot teams running research on a TV, building a domain and website, and topping at least one informal coding bench cited by Claire Vo | These are individual demos, not the unnamed company’s internal set |
Pros and cons
For users
- Pro: A public directory, and especially a documented set of cooperating workplace bots, would make it easier to find agents that already do a real job instead of starting from a blank prompt.
- Con: Until the company is named and the bots are visible, users cannot inspect quality, safety or whether “over 300” means 300 distinct useful tools or many near-duplicates.
For developers
- Pro: Publishing bots “in groups that work together” would give builders a rare look at how a large organization actually chains Grok Bots, which is harder to learn from isolated community listings.
- Con: An independent directory is not an official review layer. Copying internal prompts and tool setups from a corporate dump can spread brittle patterns as easily as good ones.
For companies
- Pro: Releasing internal agents can be a form of open practice-sharing, and a third-party catalog lets a firm publish without waiting on an official xAI storefront.
- Con: Internal bots often encode process, vendor names and implicit credentials. A “large company” that has not been identified cannot yet be audited for what it chose to strip out before publication.
What X is saying
Roemmele’s own posts are the primary source for both the 300-bot claim and the directory’s growth.
BOOM! A rather large company I work with has agreed to make public over 300 Grok Bots they use internally! This will be some of the most complex and useful sets of Grok Bots thus released. We will release them in groups that work together and will only be found on GrokBot.Rocks. Out soon!
— Brian Roemmele (@BrianRoemmele) November 7, 2026 · View on X
Weekly analysis is in. On schedule to break past 10,000 curated Grok Bots in the GrokBot.Rocks directory by Monday.
— Brian Roemmele (@BrianRoemmele) November 7, 2026 · View on X
holy sh*t.. my TV now runs a full Grok Bot RESEARCH team
— monokern (@monokern) October 25, 2026 · View on X
The earlier visit-count post — nearing 10,000 listings, with over 130,000 Grok Bots having visited — is the source for those two figures. They describe traffic to the directory, not the size of the promised corporate release.
Critical analysis
There are three stories here, and they should not be collapsed into one. The first is a directory: Roemmele built an independent catalog that, by his November 7 count, holds 9,587 public Grok Bots. That is checkable on his page. The second is his own published library of 36 bots. The third is a promised drop of more than 300 internal bots from a company he will not name.
Only the third item is why this topic is trending, and it is the least verified. “A rather large company I work with” is a teaser, not a disclosure. “Out soon” is a timetable without a date. “Only be found on GrokBot.Rocks” is both a distribution plan and an advertisement for the directory he runs. None of that makes the post false. It does mean the 300-bot headline should be attributed to Roemmele until the collection is public and inspectable.
The creator demos are a separate, weaker signal. A TV running a research team and a bot that registers a domain show enthusiasm and some real capability. They do not tell us whether a large company’s internal fleet is carefully governed or a pile of prompts that happened to work on one team’s accounts. Earlier i10x reporting on token limits and agent-team sprawl is the better context for that risk than a launch-week directory post.
What’s still open
- Which company is releasing the 300-plus bots, and when will the first group appear on GrokBot.Rocks?
- How much internal process, tooling and sensitive detail will be stripped out before publication?
- Will the public listing count actually cross 10,000 on the timetable Roemmele published?
- Does xAI intend an official catalog that would sit alongside, or replace, this independent directory?
About the analysis
Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated illustration. Read our AI Use Policy. Spotted an error? Request a correction.
This independent, research-based i10x analysis draws on Brian Roemmele’s November 6–7 posts, his November 7 GrokBot.Rocks page on ReadMultiplex, and public creator posts about Grok Bot tests. The company behind the 300-bot release is unnamed in the primary source and is not identified here. Related i10x coverage: Grok Bot Agent Teams: From Chief of Staff to Bot Sprawl and Grok Bot Token Limits Spark Agent Governance Shift.
- Brian Roemmele: over 300 internal Grok Bots coming to GrokBot.Rocks
- Roemmele: nearing 10,000 listings; 130,000 bots have visited
- Roemmele: on schedule to pass 10,000 curated bots by Monday
- GrokBot.Rocks page on ReadMultiplex (updated November 7, 2026)
- i10x: Grok Bot Agent Teams
- i10x: Grok Bot Token Limits
i10x Perspective
Directories become interesting the moment someone with production software chooses them as a release channel. If Roemmele’s unnamed company follows through, GrokBot.Rocks will have done in two weeks what official marketplaces often take a year to do: give working agents a public address. The caution is obvious. A teaser about “over 300” bots from “a rather large company” is a headline that can outrun the files. Watch the directory, not the announcement. The story gets real when a labeled collection appears, in groups that work together, and other people can run them.
