Is AI citing your startup? I built a free tool that tells you

Ran it on my own site immediately after reading this. Really eye-opening.

The robots.txt point is what got me, I had no idea that was potentially blocking AI crawlers. That's the kind of thing you'd never catch on a standard SEO audit.

The distinction you made between "findable" and "quotable" is the clearest way I've heard AEO explained. SEO tells you if Google can find you. This tells you if AI will actually repeat you. Those are genuinely different problems.

One thing I'd love to see added a before/after tracker so you can rescan your site after making fixes and see the delta over time. Right now it's a great diagnostic but adding progress tracking would make it a habit rather than a one-time check.

The fact that Stripe scored a C is both alarming and encouraging. If they're not optimizing for this, there's a real window for smaller founders to get ahead of them on niche queries.

The SEO vs AEO distinction is something more tool builders need to internalize. I've noticed the same gap with my own free utility site — traditional SEO signals look fine, but getting cited by AI requires a completely different content structure.

The key insight from your post: LLMs need something verbatim-liftable. A punchy H1 slogan fails AEO even if it ranks on Google. The sites that are getting cited have a dense, self-contained "what is X" paragraph that answers the question before anything else. That's the structure to optimize for.

The llms.txt finding is also underrated. Most sites don't even know they need one. Adding FAQ schema + llms.txt in the same afternoon is probably the highest ROI 2-hour session a founder can do right now.

just ran mailtest.scuton.com through it — 50/100, grade D. ouch.the surprise hit: robots.txt is blocking GPTBot AND ClaudeBot entirely. i'm building an email deliverability tool and my own site is invisible to the two AI systems most likely to surface it. classic.also 0 of 23 headings in question format. i thought SEO meant "don't write questiony headers, keep them punchy." turns out that's exactly the wrong move for AEO.fixing the robots.txt and llms.txt tonight. the question-format headings one is going to take longer. appreciate

this — useful in a way most free tools aren't.

I just ran my web app through it and got a 56/100, C. Humbling but fair.

What surprised me most: FAQPage schema, llms.txt, and all three AI bots scored 100. I literally just added the schema two days ago after a separate audit flagged my AEO score at 48/100. So the tool caught the improvement which is a good sign it's actually reading live data.

The H1 and canonical URL findings are legit and actionable. Going to fix those today.

The low content depth score is tricky for SPAs, my landing page is React-rendered so static crawlers only see ~5 words of pre-JS HTML. Not sure if that's a tool limitation or something worth solving with SSR/prerendering.

One bug worth flagging: I signed up with my email to unlock the full report including the 5 live AI prompts, but after logging in it's showing the same "sign up to unlock" CTA instead of the full results. Seems like the authenticated state isn't being recognized properly after signup → login flow. Might be worth checking the post-signup redirect.

But overall, this is a genuinely useful tool! The live prompt testing against real AI engines is the differentiator. Everything else you can cobble together from free audits but that part you can't. 🤝

GEO is the Wild West right now. Two questions on your tool. First, how do you source the "is AI citing you" signal. Are you actually querying GPT, Claude, and Perplexity with test prompts and parsing results, or inferring from something else like brand mention counts?
Second, do you see dramatic variance by platform? Anecdotally Perplexity cites small indie products way more often than ChatGPT does in my testing, probably because of more aggressive default web search. Would love to see a per-platform breakdown in the tool.

The shift from being “searchable” to “quotable” feels bigger than most people are admitting. A lot of sites still read like branding pages, not answers.

Also that Stripe and Vercel part is kind of wild. Makes it feel less like a resource problem and more like a blind spot.

The robots.txt point is probably catching more people than they think too

Feels like the phrasing layer might matter more than the technical fixes right now

This is exactly the problem I've been thinking about. Traditional SEO and AI citation optimization (AEO/GEO) are diverging fast. Your product getting ranked on Google doesn't mean ChatGPT or Claude will recommend it.

I've been working on something related — curating reusable AI coding assets that are structured to be easily cited by LLMs. Check out tokrepo.com/en/resources/59436371-30d6-4a51-9f9b-1b1986873728 for an example of how we compare AI coding agents in a format that's both human-readable and AI-quotable.

The "quotability" angle is underrated. Most founders still optimize for clicks, not for being the answer an AI gives.

The SEO vs AEO distinction you are making here is one of the more practically useful framings I have seen in a while. We noticed the same thing building ActComply (EU AI Act compliance tool): ranking well on Google for "EU AI Act compliance" does not mean the product gets recommended when someone asks ChatGPT or Perplexity "how do I check if my AI system is compliant."

The signals that seem to matter for AI citation are different: unambiguous factual claims, clear entity definition (what the product is, what it does, who it is for), and being referenced from sources that AI training sets trust. Which means press, research institution mentions, and authoritative community posts like IH actually matter more for AEO than they do for traditional SEO where raw backlink volume dominates.

Your point about being "quotable" is the key one. A landing page optimized for conversion is not optimized for quotability. Those two goals actively conflict in ways most founders have not thought about yet.

Feels like most sites are still writing for SEO-era patterns — long intros, storytelling, then the answer buried somewhere in the middle. LLMs don’t care about that flow.

They just want something they can lift cleanly.

The FAQ schema point is interesting too. Makes me think structured content will dominate citations going forward.

Curious — have you seen any pattern in how tone affects citations?

This is a great product! I had considered GEO before, but your product makes everything ever clear. It's perfect for new developers, and I can't wait to try it out.

good

The shift from traditional SEO to Answer Engine Optimization (AEO) is a wake-up call for everyone who thinks a "green" Lighthouse score is enough to stay relevant in 2026. You’ve pinpointed the exact reason why even industry giants like Stripe are failing to be cited: they are optimized for discovery by humans, but not for synthesis by Large Language Models. To be "quotable" by an AI, a site must move beyond catchy slogans and focus on "Information Gain," using structural tools like llms.txt files and FAQPage schema that act as a direct roadmap for crawlers like GPTBot and ClaudeBot. Your audit reveals that the future of digital visibility depends on being "synthesizable"—creating self-contained, fact-rich answer blocks that an AI can lift verbatim. For those looking to see how these principles apply to high-performance mobile platforms and specialized app optimization, you can find practical examples of structured content here modwinkapk.

The GEO gap you've found is massive, but as SaaSOffers scales, the real danger isn't the tech, it's the 'Founder's Trap.' You’re about to hit the wall where managing the feedback loops and roadmap eats the time you should be spending on the vision.

I’ve helped scale similar workflows by taking the operational weight off the founder's plate. I’m currently looking for one high-conviction project to help professionalize for Q2.

Not sure if you’re ready to let go of the 'Project Manager' hat yet to focus purely on the build, but if the growth is getting chaotic, we should talk. I’m only taking on one more partner this month before I’m locked in.

This is a really interesting shift. The idea that being “quotable” matters more than just being indexed feels like a big change.

It also explains why some smaller tools are suddenly showing up in AI answers while bigger names are missing.

One thing I have been noticing alongside this – even if you do get the content right, distribution is still a separate problem. A lot of solid tools just never get seen early enough to even be “in the mix.”

I have been experimenting with a simple feed where you can post your project and get it in front of other builders without needing an audience first: https://buildfeed.co

Might be worth dropping your tool there as well, especially since this is the kind of thing founders would actually try and give feedback on.

Curious what scores others here are getting too.

This is really interesting it feels like we’re moving from “can AI find you” to “how does AI use you.”

Curious if you’ve thought about what happens when tools start acting on that information (not just surfacing it)?

I’ve been seeing cases where the output is technically correct, but the downstream action it triggers is where things break.

This is exactly the problem I've been solving. I built App — an AI code auditing tool that runs automated security and quality scans and delivers a professional PDF report ranked by risk. Happy to run a free audit for anyone here who wants to see what it finds in their codebase.

Thank you for all this information! I've got my first product launch coming up this week and didn't know the first thing about AEO. In didn't even know it was it's own thing! It's 3am where I am now, but I've and left myself a reminder to check this out in the morning. Thank you!

WOW, the brave new world of AI leaves us all speechless... Good luck with the project...

Ran it on ReviewLift. Got 58/100 (Grade C). Ouch.

The biggest gap: zero structured data. No FAQ schema, no Organization schema, no nothing. The tool says adding FAQPage schema alone is the #1 lever for AI citations.

Also learned my Open Graph tags are completely missing—so links shared on X/LinkedIn look terrible. That's an easy fix I just never checked.

This is a really interesting angle — especially the idea that “being quotable” is now more important than just being searchable.

One thing I’m curious about: how stable are the results from the prompt testing side? If you run the same 5 prompts a few times (or tweak wording slightly), do you see big swings in whether a brand gets cited, or is it fairly consistent?

Trying to figure out whether this is something you can reliably optimise for, or if there’s a lot of randomness in how LLMs decide what to include.

great idea

Debt collection is one of the most searched legal topics in the country and almost none of the content ranking for it can actually be cited by AI because it's written by content farms hedging every sentence. The first person to publish clean, direct, citable answers to how the FDCPA actually works is going to own that category in AI search. The window is open right now.

this really resonates. I have been seeing something similar with my AI tool — traditional SEO signals can look healthy, but that still doesn’t mean AI will mention you when someone asks a real buying-intent question.

Nice that you tested this on real prompts instead of stopping at technical checks. That feels much closer to how products are actually discovered now.

Read you post and instantly tried the same rodeo with ChatGPT, turns out you're right. Thanks for sharing!

the gap between SEO rank and AI citation is something I keep running into - Google and LLMs have completely different data pipelines. does the tool show per-model breakdown? that'd be the most actionable part.

Good post and tool. Didn't know AEO was yet a thing before reading, thanks for sharing.

This hits so close to home, honestly. I’ve been spending way too much time digging through Twitter, Indie Hackers, and random forums just to find relevant SaaS offers and discounts — it’s such a scattered mess right now. Your tool feels like exactly what the community’s been missing: one single place to track, compare, and actually use these deals without jumping through 10 different links. Love that you’re solving a real, annoying pain point instead of just building another “me-too” AI tool. Curious though — do you have plans to add filters for specific niches (like B2B vs. creator tools) or price tiers? That’d make it way easier to find stuff that fits different budgets/use cases.

Good insight

Good to read

Great one. I learned a lot.

The FAQ schema insight is crucial. Search quality is primarily about content structure and clarity, not model sophistication. Most sites treat their content as afterthoughts for search engines rather than designing it to answer real questions first. FAQ schema forces that discipline. This applies to any semantic matching system.

Good read. The "Stripe got a C" moment is the strongest hook — most founders assume big companies have this figured out, and that frame alone will push people to scan.

Couple of honest reactions:

1. The FAQ schema → 3-4x citation lift is the non-obvious finding. If that's real at scale, it's a cheap win most sites can ship in a day.

2. The "robots.txt blocking GPTBot by default from a WordPress plugin" pattern — brutal, and I bet more common than people realize. Worth a standalone post just on that.

3. I don't have a site up yet for my own thing (Day 3 of building SubKitt, an AI agent that turns technical founders' shipped work into distribution), but I'm going to keep this bookmarked for when I do. AEO is probably underweighted in the build-in-public community relative to how much it'll matter in 12 months.

When I run it, I'll report back with the score + which fix surprised me. Fair?

The live prompt testing is the killer feature here. Audits telling you “technically fine” don’t matter if the model still doesn’t mention you. Curious if you saw big differences between brand mentions vs. direct quoting of content.

Is AI citing your startup?

I developed a free tool that helps you instantly check whether AI systems are mentioning or citing your startup in their responses.

smartsimreg

Read more → Check your AI visibility

Ran this on one of the blogs I’m working with and I got a 44 (Grade D) 😅

The biggest gap was exactly what you mentioned . The content was sitting inside a Google Docs layer, so there’s nothing for LLMs to actually parse or quote.

As a SaaS content writer, this is a big realization. We’re not just writing for ranking anymore; we’re writing for extraction.

Going to move this to a proper HTML page + add structured FAQ blocks and test again. Curious to see how much the score improves.

Good insight but citation is a proxy.
the real metric is: does it drive pipeline?
AEO alone won’t hold as an edge for long.

Totally, that move from "be findable" to "be quotable" is really huge. ~

Search was just about getting them to your site. This is all about getting them to lift words out of your site. Completely different motivation.

And exactly, once it’s trimmed down to a direct, clean answer, it’s automatically shareable and reusable. Wrapped content will always just be ignored.

Your strategy is smart. A clean, self-contained answer, and then wrap around it-that seems like the simplest way without blowing up your existing content.

On the convergence idea, it's already happening, to a degree. A lot of pages are migrating to this "lean answer + fluffy content" style.

But if everybody does it, then formatting really won't matter anymore.

Then it becomes about:

who is trustworthy

who gets cited more

who has been correct more often

It's going to be similar to what we saw with SEO. Initially, it was about structure and keywords. Later, it became about authority.

It's a gradual thing at the moment, but we can already see that some publishers are being cited again and again, even when others say the exact same thing. That’s where this will probably go as well.

So formatting gets you to play the game, but credibility will be key to winning it.

Interesting — will try this on my site. The "quotable not findable" framing is spot on.

This is interesting — especially how Reddit played a role here.

Feels like many tools focus heavily on analysis, but the real challenge is turning that into actual visibility and leads.

Combining both in one flow makes a big difference.

Ran it on my own site. Got a humbling score. The two fixes that actually changed my numbers: the FAQ schema (your 3-4x claim is real — first query I tested went from 0 mentions to appearing in 2/3 responses after adding it) and cleaning up the robots.txt.

The robots.txt finding is the sneakiest one in here. Founders don't know their WordPress security plugin silently blocked ClaudeBot. It's not even malicious — it's just that "block all bots" was the safe default for a decade and nobody updated the mental model when LLM crawlers arrived.

The "quotable not findable" framing is the sharpest line I've seen on this topic. SEO people keep asking "but does this affect my Google rankings?" — completely missing that they're different distribution channels now. Getting cited in a ChatGPT response and ranking #3 on Google are unrelated outcomes that require different inputs.

One question: how do you handle variance between models? In my testing, Claude and ChatGPT will sometimes give opposite answers on "best X for Y" — Claude cites one brand, ChatGPT cites a competitor. Do you average across models or show them separately?

Tried this on my site and honestly same experience — SEO looked fine but AI barely mentioned me. The answer paragraph + FAQ schema fix alone made a noticeable difference.

This comment was deleted 4 months ago