Ten minutes is enough to choose a useful starting point from a Grok Bot directory. It isn't enough to inspect every bot. Reading every listing and testing each one against your own notes takes an afternoon. The goal is narrower: identify one template aimed at the job in front of you, check the public signals around it, and open it with a concrete first task.

We organize public bots by job, not by vague claims about intelligence. As of September 7, 2026, the official xAI marketplace listed 69 public bots; mapped into our categories that is 35 in Sales & marketing, 12 in Personal assistant, 11 in Other, 6 in Operations & admin, and 5 in Developer & data. Those counts tell you where the broad selection is. They don't tell you which bot is best for your work.

The useful signals come next. A vote shows how many signed-in members backed a listing. An add counts an open through the listing link. Creator credit identifies who made the bot and points back to the public source. An add isn't an install or download because xAI doesn't publish install counts. Together, votes and adds show community interest and intent. None of these signals is a guarantee of results, but each is more concrete than a polished name.

Minute 1: write the job as an output

Start with one sentence that names what you need returned. Write “turn these meeting notes into owners and deadlines,” not “help with projects.” Write “find gaps in this landing page for search,” not “improve marketing.” A specific output gives you a way to reject close-looking templates quickly.

Include the material you already have and the form you want back. A useful sentence has three parts: the input, the action, and the output. For example: “Use these call notes, identify the open commitments, and return a follow-up list with owners.” This takes less than a minute and stops category browsing from becoming aimless.

Minutes 2 and 3: choose the closest category

Match the noun and verb in your sentence to the directory. Work involving outreach, posts, clips, SEO, or leads belongs in Sales & marketing. Code, reviews, and technical research belong in Developer & data. Staffing, schedules, and internal briefs point to Operations & admin. Inbox, calendar, home, and individual routines point to Personal assistant.

Use Other only when the job genuinely doesn't fit a named category. A larger category isn't automatically a better category. Sales & marketing is many times larger than Developer & data in the current set. The smaller list can still be the faster route when your desired output is technical.

Don't switch categories because a bot name sounds clever. Category is the first filter because it reflects the job. The name is only a label.

Minutes 4 and 5: rank by votes, then check adds

Within the category, start with the vote order. Votes are a direct community action in the directory, so they provide a consistent first pass across listings. Open the leading candidates in separate tabs, then compare adds.

Votes and adds answer different questions. A vote says a signed-in member chose to back the listing. An add says someone opened the bot from the directory. A listing with votes and adds has earned two forms of attention. A listing with fewer votes but repeated adds can still deserve a look because people are moving from the directory into the bot.

Don't turn either number into a quality score. We don't test, audit, or certify the bots. The ranking is a navigation tool. It helps you decide what to inspect first when several listings target the same job.

If two candidates remain close, choose the one whose description names the output in your sentence. A bot aimed at project follow-through is a closer match for owners and deadlines than a general writing bot, even if the general bot has more activity elsewhere.

Minutes 6 and 7: follow the creator credit

Every useful listing should preserve creator credit. Read the handle, follow the source link, and compare it with the creator named on the public x.ai page. This gives you provenance: a person or team publicly attached to the template.

We keep the creator's handle and the source link on every listing so credit stays with the person who did the work, and so you can check the bot at its origin.

For example, the directory credits Projects Manager to @ericzakariasson, and its public x.ai bot page carries the template itself. The current source set also includes SEO & AEO Desk by @adamta and Researchy by @farzyness. These are facts you can check at the source rather than endorsements written by the directory.

Creator credit is a trust signal because it lets you inspect origin and context. It isn't a performance badge. A familiar handle doesn't prove that a bot fits your files, permissions, or standards. An unfamiliar handle doesn't make a template unusable. Credit simply gives you an accountable starting point and preserves the work of the person who shared it.

Reject a candidate if the listing and source don't line up. Names can change, and similar bots can exist. The public bot URL and creator context are stronger identifiers than a title alone.

Minutes 8 and 9: read the bot page against your sentence

Open the public x.ai page for each finalist. Check that the bot's stated purpose matches your requested action and output. Keep your review tied to the job sentence. Features that don't affect the first task aren't deciding factors.

Choose the narrower fit when the first task matters more than future flexibility. Projects Manager returns owners and deadlines from notes. SEO & AEO Desk finds gaps in a page for search; Researchy returns a sourced research brief. Their names, credited creators, categories, and public pages give you enough information to choose which one deserves the first trial without inventing capabilities beyond the source.

Before opening a bot, remove private material from your trial input. Use a short, representative example. The first run should answer one question: does this template return the kind of output you asked for?

Minute 10: make one reversible choice

Pick one candidate. Record its listing URL, vote count, add count, creator handle, and the job sentence you wrote. Then open the bot through the listing and run the small example. This creates a clean record of why you chose it and makes a later comparison straightforward.

If the output misses the requested form, return to the other finalist. You haven't committed to a platform migration or a long setup. You have completed a ten-minute selection pass using job fit, community ranking, open activity, and source credit.

The method stays simple on purpose. Category narrows the field and votes order what is left. Adds and creator credit tell you which of those to open first. Start now by opening the Sales & marketing directory and write your one-sentence job before choosing a bot.