Free AI Visibility Tools: Brand Report, AI SEO Audit, Query Fan-Out Generator, and Crawler Inspector

You can learn a lot about AI visibility without a contract. The catch is knowing which free thing you opened. Promptwatch sells three different "free" surfaces: a public tools page, a free in-app tracker called Explore, and the paid platform you graduate into. Mixing those up is how people say they "already have Promptwatch" after running one audit, and it is also how a team builds a program on a snapshot that does not refresh. This post separates the three so the free tools get used for what they do and the paid platform gets bought for what it does.

The mixing is the part that causes the most downstream confusion, because the three surfaces share a brand and a login flow and not much else. A team that runs the Brand Report once and tells procurement they "have Promptwatch" has a snapshot, not a platform. A team that opens Explore and treats it as the monitoring program has ten prompts on one engine, not a program. The three surfaces are different products with different jobs, and the job each one does is the thing that decides which one a team should be in. The separation below is the one we give to buyers who ask which free thing they opened.

Google's AI features guidance still covers Overviews. It does not store ChatGPT or Perplexity answers on a prompt you typed, which is the gap the rest of this page fills.

The Google doc is the one piece of the stack that is free and official and covers one surface. AI Overviews are a Google surface, and Google documents them. ChatGPT and Perplexity are not Google surfaces, and Google does not document them. The gap between what Google documents and what a team needs to measure is the gap the rest of this page fills, because the gap is the non-Google surfaces, and the non-Google surfaces are the ones a team has to measure with a tool.

The public suite lives at promptwatch.com/free-tools. Promptwatch the product is the dashboard behind the login. Affiliate link if you are ready for the app: promptwatch.com.

The public tools

The AI Brand Visibility Report is a one-shot look at visibility scores, sentiment, and competitors across models. Useful as a first read. It is not a daily prompt ledger, so it answers "where do we stand" and not "what changed since last week." A team that runs it once and treats the score as a baseline has a baseline, not a program.

The Brand Report is the one that gets mistaken for a program most often, because it produces a score and a score feels like a measurement. The score is a point-in-time read, which is a baseline. A program is a time series, which is a score that moves and can be compared to last week's score. The team that runs the report once and stops has a baseline and no way to see movement, which is the difference between a snapshot and a program. The report is the right tool for the baseline and the wrong tool for the program, and the distinction is the one that keeps the team from building on a snapshot.

The AI SEO Audit is a site pass aimed at ChatGPT, Claude, Perplexity, and Gemini readiness. Treat the output as a punch list, not as Agent Analytics. The audit tells you what to fix on the page. It does not tell you whether the bot that should fetch the page did, because that is a log feature rather than an audit feature.

The audit is the one that gets mistaken for crawler logs, because both tell you something is wrong with a page. The audit tells you what is wrong with the page content, which is a punch list of fixes a content team can ship. The log tells you whether the bot fetched the page, which is a fetch record a technical team can act on. The two are different diagnostics for different problems, and the audit is the content diagnostic and the log is the fetch diagnostic. A team that treats the audit as a log diagnostic ships content fixes for pages the bot cannot reach, which is the wasted work the log is meant to prevent.

The ChatGPT Query Fan-Out Generator shows how ChatGPT expands a prompt into extra searches. Use it to see the hidden queries that sit behind the question you typed. It will not store 50 prompts and refresh them tomorrow, so it is a thinking tool for prompt design rather than a monitoring tool for prompt tracking.

The Fan-Out Generator is the one that changes how a team writes prompts, because it shows the cluster the engine reads instead of the keyword the team typed. A team that sees the fan-out writes pages for the sub-questions, which is the work that produces citations. A team that does not see the fan-out writes pages for the head term, which is the work that produces paraphrases without citations. The generator is a thinking tool, not a monitoring tool, which means it runs once per prompt design session and does not run on a schedule. The schedule is the monitoring tool's job, and the generator is the design tool's job.

The Website LLMs.txt Generator builds an llms.txt file. Do not claim Google uses it. Promptwatch's own research says the file has little or no impact on AI search. It is optional documentation for agents that opt in, not an Overview hack, and a slide that promises rankings from the file is a slide that will not survive a follow-up question.

The llms.txt generator is the one that gets oversold most often, because the file got attention and the attention produced claims the file does not back. The honest framing is that the file is documentation for agents that opt in to read it, which is a narrow use case and not a ranking mechanism. A slide that promises rankings from the file is a slide that falls apart the first time a stakeholder asks for the evidence, because the evidence is the research that says the file has little or no impact. The generator is the right tool for teams that want the documentation and the wrong tool for teams that want a ranking hack.

The AI Crawler URL Inspector shows the extracted markdown and metadata a model is likely to see, which helps when the page is a JavaScript shell that returns nothing useful to a bot. The Page Token Inspector shows how a model tokenizes the page, which helps when the template is bloated and the token budget is going to chrome instead of content. Both are diagnostic tools for one URL at a time.

The two inspectors are the ones that diagnose the page-level problems the audit cannot, because the audit works on the page content and the inspectors work on what the bot sees. The Crawler URL Inspector is the one that catches the JavaScript shell, which is a page that renders for a human and returns nothing for a bot, and the catch is the one that explains why a well-written page gets no citations. The Page Token Inspector is the one that catches the bloated template, which is a page where the token budget goes to navigation and chrome instead of the answer, and the catch is the one that explains why a long page gets a short citation. Both are one-URL-at-a-time tools, which is the right scope for a diagnostic and the wrong scope for a program.

The robots.txt generator for AI writes starting allow and deny rules. Read the file before you ship it, because a generated block rule can take SearchBot out along with the training bots. The AI Bots directory lists 400-plus crawlers with operator, purpose, and robots.txt notes, which is what you reach for when a mystery UA shows up in your CDN logs. The glossary covers 250-plus GEO, SEO, and AEO terms if a stakeholder still says "ChatGPT SEO" for everything.

The robots.txt generator is the one that produces the most accidental blocks, because a generated deny rule can match more than the team intended, and a deny that takes out SearchBot is a deny that takes out the citations. The read-before-ship step is the one that catches the accidental block, and the step is the one that turns the generator from a risk into a time-saver. The bots directory is the one that demystifies the CDN log, because a mystery user agent in the log is a question the directory answers, and the answer is the one that decides whether to allow or block. The glossary is the one that fixes the language, because a stakeholder who says "ChatGPT SEO" for everything is a stakeholder who conflates the surfaces, and the glossary is the one that gives the team the words to separate them.

There is also a free tool idea generator and a meta image previewer on the same page. Those are extras, not monitoring, and they do not belong in a measurement plan.

Explore is a different product

Explore is the free in-app tracker: ChatGPT only, 10 prompts, 1 project, 1 seat. That is a sandbox for Visibility Score on a handful of strings. It is not the Brand Report, which is a public snapshot. It is not crawler logs, citations by type, Content Agents publishing to Webflow, or MCP. A team that needs any of those has outgrown Explore on the first day, and that is the correct outcome rather than a failure of the free tier.

Explore is the one that gets mistaken for the platform, because it lives inside the app and the app is the platform. The difference is that Explore is a sandbox for ten prompts on one engine, and the platform is a program for many prompts on many engines with the logs and the citations and the publishing. A team that needs the sandbox gets it for free, which is the right outcome. A team that needs the program outgrows Explore on the first day, which is also the right outcome, because the free tier is not meant to be the program. The outgrow is the signal that the team is ready for the platform, and the signal is the part that makes the free tier honest.

The paid platform starts at Essential, $95 a month with a 7-day trial, covering 50 prompts, 6,000 responses, daily refresh, multi-engine UI checks, and MCP and API. Professional is $245 a month. Agency Kick-off is $199 a month. Promptwatch is rated 4.7 out of 5 on G2 across 1,840-plus brands.

Otterly.AI from $29 a month is a paid monitor with a no-card trial, not this free-tools page. Peec AI from $95 a month is paid. Profound Starter is $99 a month annual. Method: how we rank. Tools list.

FAQ

Is the Brand Visibility Report the same as Explore?

No. The report is a public snapshot. Explore is 10 saved ChatGPT prompts inside the app.

Should we ship llms.txt from the free generator?

Only if you want a readable map for agents that opt in. Do not expect Google or ChatGPT rankings to move because the file exists.

What to do this week

Run the Brand Visibility Report and the AI SEO Audit on the production domain.
Put three money prompts through the Query Fan-Out Generator and list missing pages.
Inspect one JS-heavy URL in the Crawler URL Inspector.
Open Explore and save 10 prompts if you want a repeating ChatGPT check.
If you need Gemini, citations, or logs, start Essential. Do not stretch the free tools page into a program.