AIO Checker
AI crawler readability audits for SaaS websites
I built a small tool called AIO Checker. The first version got 3 real payments after an accidental Indie Hackers spike: two at $4.99 and one at $1.99.
Then it went quiet.
The mistake was not checkout or pricing. It was distribution, and also positioning.
The early version was too centered on llms.txt. After rebuilding it, I think the more useful question is:
Can AI/search crawlers actually fetch, parse and understand your public site?
So the new version checks boring but practical things:
- robots.txt and AI crawler rules
- indexability
- readable initial HTML
- sitemap discovery
- structured data
- snippet controls
- optional llms.txt
The free scan works without login. The paid part is a one-time $19 ZIP with fix.md, robots/schema snippets and prompts for Cursor/Claude/ChatGPT.
I am trying to validate whether this should stay as a one-time report or become an agency/multi-site pack.
Try it:
If you run a SaaS or devtool site, post your score. I am especially interested in which stack patterns fail hardest.
I’ve been trying to build on the internet for about 6 months now.
Not in the “I want passive income while I sleep” way.
More like: I want to build enough leverage that I don’t have to spend my life working on things I don’t believe in. I don’t want to look back one day and realize I was useful to payroll systems, Jira boards, and other people’s roadmaps, but never built anything meaningful of my own.
The uncomfortable part is that building is not the hard part anymore.
With Cursor, Claude, ChatGPT, Vercel, Stripe, Supabase, etc., I can ship faster than ever.
The hard part is not disappearing.
I had this realization after launching a small tool and checking Google Search Console.
The site existed.
The pages worked.
The product was usable.
But Google had basically only noticed the homepage and one old redirected page.
Everything else — rankings, example report, guides, privacy, terms — existed, but might as well have been invisible.
That hit me harder than any failed launch.
Because as indie hackers, we obsess over:
- features
- design
- pricing
- stack
- “should I add auth?”
- “should I add Stripe?”
- “should I rebuild this in Next?”
But then we ship a site that search engines barely understand and AI crawlers may not be able to read.
So I rebuilt my project around that problem.
The product is called AIO Checker:
It checks whether a public site is actually readable as a source for search and AI systems:
- crawl access
- indexability
- initial HTML readability
- structured data
- sitemap quality
- llms.txt detection
- AI crawler rules
- snippet / exposure controls
I ran it on itself after the rebuild:
- AI Visibility: 92/100
- AI Exposure: 49/100
- 591 visible words in initial HTML
- 27 sitemap URLs
- JSON-LD found
- llms.txt found
The funny part: the tool also exposed its own weaknesses.
It told me I still have thin trust signals and that my crawler exposure policy is too open if I care about training bots.
That felt honest, which is what I wanted.
My current thesis:
Most indie products don’t die because the product is bad.
They die because the internet never learns how to describe them.
Google doesn’t index enough pages.
AI systems can’t parse the React shell.
The homepage is vague.
The sitemap is incomplete.
The founder writes one launch post, gets 11 upvotes, then assumes the market rejected the product.
Maybe the market never even saw it.
So my next bet is simple:
Create useful, boring, indexable pages around real search intent.
Not “AI SEO ultimate guide 2026” garbage.
Actual pages like:
- why ChatGPT can’t read my React site
- how to allow ChatGPT Search but block AI training
- do I need llms.txt?
- AIO Checker vs traditional SEO audit
- public scans for Vercel, Stripe, GitHub, etc.
I don’t know if this will work yet.
But I know the previous version — building quietly and hoping people magically arrived — definitely didn’t.
If you’re building something right now, I’d be curious:
When was the last time you checked whether your site is actually readable before JavaScript runs?
Drop your URL if you want. I’ll run a few through AIO Checker and reply with the most obvious issue I see.
-
1
Small update:
I’m going to run a few free scans from this thread and reply with the single most obvious visibility issue I find.
Not a full audit, just the highest-leverage thing I’d fix first.
If you drop your URL, I’ll look at things like:
whether the initial HTML is actually readable
whether the sitemap exposes useful pages
whether robots.txt blocks or allows the right crawlers
whether structured data exists
whether the homepage explains the product clearly enough for search / AI systems
I’m especially curious to see how many indie products are technically live, but invisible before JavaScript runs.
I built AIO Checker a while ago and it even got a few real payments.
But the first version had a problem: it treated llms.txt as too central.
After reading the current docs from Google/OpenAI/Anthropic and looking at how AI crawlers are actually documented, I rebuilt the product around two different jobs:
1. Be visible: can AI/search systems crawl, parse and understand the site?
2. Stay protected: are AI crawlers allowed to access things the owner may want controlled?
The new V3 has:
- free scan
- no login
- AI Visibility Score
- AI Exposure Score
- public example scans
- full ZIP report
- fix.md for Cursor/Claude/ChatGPT
I also raised the paid report from "cheap impulse" pricing to $19 because the old price made the product feel less serious.
Would love feedback on the positioning:
https://aiochecker.vercel.app
Hey Indie Hackers,
I launched AIO Checker a few days ago – a tool that audits websites for AI visibility (ChatGPT, Claude, Perplexity, etc.). My first "launch" was on Hacker News. Got ~100 visits, $0 revenue.
The problem?
I was giving away the entire solution for free. My initial UI showed all the recommendations and code snippets right there in the network tab and on the screen. No reason to pay.
So I did a hard pivot in 24 hours:
Strict Paywall + Blur – Now the detailed fixes are blurred until payment or unlock.
Feedback‑for‑Unlock – The first 20 users can unlock the full report by leaving a genuine review. This gives me social proof and them a free report.
Social Proof Wall – Live on the homepage, showing real feedback from early users (already got the first one!).
Result so far:
First feedback received and displayed ✅
Site feels way more trustworthy
Learned that traffic ≠ revenue – you need to gate value correctly
Why this approach?
Free users get value (score + basic metrics), power users pay for deep insights and copy‑paste code snippets.
The 20 free slots create urgency and bootstrap testimonials.
The social proof wall builds credibility for future visitors.
What I'm still figuring out:
How to get that first organic paying customer.
Whether the $4.99 price is too low/high for impulse buys.
How to reach more site owners (not just makers).
I'd love your feedback:
Does the new flow make you want to try it?
Any suggestions to improve conversion or the unlock mechanism?
What's your experience with launching and pivoting quickly?
Check it out: https://aiochecker.vercel.app
Thanks for reading! 🙏
#buildinpublic #microsaas #ai #seo
-
2
This is a great post because early feedback is such a hard step for most founders.
I also think social proof (even small proof) can change how people respond to a new product.
What was the single launch mistake that hurt the most at first — and what change fixed it fastest?
-
2
hope it works out well for you congratulations
-
2
The $0 → first feedback transition is the hardest part, and you nailed the frame: it's not a revenue problem, it's an evidence problem. Nobody wants to be first.
Question for you: what was the actual "launch mistake" you fixed in those 24 hours? You mention you fixed it but I couldn't tell from the post whether it was a positioning change, a pricing change, a CTA change, or something else entirely.
I'm in the same position right now — 7 Gumroad products live, $0, trying to figure out which of those levers is actually broken. The social proof wall idea is smart because it lowers the trust barrier without changing the price. Did you find that even a few testimonials from free/beta users moved the needle?
Also: the AI visibility angle is clever. Most SEO content is still written for Google, not for how Claude/ChatGPT actually surface recommendations. That's a real gap.
-
1
You nailed it—it’s an evidence problem.
The 'launch mistake' was zero friction. I gave the full actionable audit (the actual code fixes) right on the screen. There was no reason to pay. The fix was blurring the details and adding the feedback-for-unlock loop. It forced engagement and built the social proof wall instantly.
Regarding your 7 Gumroad products: traffic without sales usually means a trust or perceived value gap. Have you tried a similar 'pay with a review' beta phase for one of them? Happy to take a look at your links if you want a second pair of eyes.
-
-
2
the feedback-for-unlock mechanic is smart. youre essentially paying for testimonials with product access instead of cash, which bootstraps social proof and gives you real user data at the same time.
the 100 visits $0 revenue learning is one everyone needs to internalize. had the same experience - gave away too much for free on my AI video tools early on. the fix wasnt just adding a paywall though, it was understanding which specific piece of output users would pay for. in your case thats the actionable fix list, not the score itself.
one thing to watch: the 20 free slots create urgency but also cap your testimonial collection. consider keeping that loop open longer - every new testimonial compounds your conversion rate.
-
1
That’s a brilliant point on the compounding effect of testimonials. I initially capped it at 20 just to force some early urgency, but you're absolutely right—capping the social proof is counterproductive in the long run. I'm going to leave that loop open or create a hybrid tier for it.
Glad to hear I wasn't the only one making the 'give it all away for free' mistake early on. Did you end up gating specific features on your AI video tool, or just the final output?
-
1
This comment was deleted 7 months ago
-
-
2
Love this — most founders underestimate how powerful early social proof is.
Building a public feedback loop in 24h is a smart move.
Curious — did showcasing the feedback immediately change conversions or just engagement at first?
Hey hackers,
I was experimenting with AI crawlers recently and got hit with a harsh reality check: most modern SPAs (React, Next.js) are practically invisible to bots like ChatGPT, Claude, and Gemini.
Because AI bots don't execute JavaScript well, if your site lacks proper SSR, they often just see an empty body tag.
I tested my own company's website expecting a decent result. It had no /llms.txt, no JSON-LD structured data, and pure client-side rendering. It scored a 24/100. 💀
I realized standard SEO isn't enough anymore. We need AIO (AI Optimization). So I spent a few nights building AIO Checker.
Here is how it works:
1. You drop your URL.
2. It scans for 7 key AI-visibility factors (robots.txt permissions, llms.txt, SSR, etc.).
3. It gives you an instant 0-100 score.
The feature I built for myself:
I hate fixing these things manually. So I added a feature that exports a detailed .md audit file. You literally just download it, paste it into Cursor, Windsurf, or Claude, and tell the AI to "fix my codebase based on this audit."
The initial scan is free, and the full export is $4.99.
Would love for you guys to test your own sites. I bet a lot of you will be surprised by what ChatGPT actually sees when it tries to read your landing pages.
Let me know what you score, and please roast the UI/UX so I can improve it!
-
2
this is a real gap. been dealing with the same thing building openslop.ai - our docs and landing pages were essentially invisible to AI assistants because of client-side rendering. the fix was straightforward technically (SSR + structured data) but most founders dont even know to check for it.
the audit-to-cursor pipeline is clever. thats basically the workflow every dev is already using - find problem, paste into AI, fix. you just formalized it.
one suggestion on distribution: instead of selling individual scans, consider making the free scan more shareable. let people embed their score badge on their site or share it on twitter. that creates a viral loop where every scan becomes marketing for you. the $4.99 price point is good for impulse buys but discoverability is your real bottleneck right now.
-
1
The embeddable badge idea is brilliant for distribution. That's a perfect viral loop, definitely adding it to the roadmap.
Just checked out openslop(.)ai, really cool concept. Did you end up implementing SSR + structured data across all your docs, or just the main landing pages? If you want to see exactly how Claude/ChatGPT are reading your setup right now, run it through the checker. I'd love your technical feedback on the
.mdcode fixes it generates for you.
-
-
2
This is a real problem that most devs don't even realize they have. I ran into something similar last year — had a Next.js app with app router and everything was client components. Googlebot handled it fine because it executes JS, but when I checked what ChatGPT was actually seeing via its browsing feature, it was basically pulling from cached snippets and hallucinating the rest.
The llms.txt angle is smart. I've been seeing more discussion about it but almost nobody has actually implemented it yet. It's one of those things where being early matters — if AI chatbots start recommending tools based on what they can actually parse, sites with proper AIO will have a massive advantage.
On pricing — $4.99 feels right for an impulse buy, but I'd consider whether a one-time scan is the right model. Sites change constantly, and AI crawler behavior evolves too. A monthly check-up subscription (even at like $3/mo) might retain better than one-off purchases. Just a thought.
-
1
Spot on. Googlebot executing JS gave us all a false sense of security for years. The AI crawlers are way more brutal with SPAs. The llms.txt standard is definitely going to be the new robots.txt.
A monthly subscription for continuous monitoring is a fantastic idea for phase 2. Right now, I just want to validate the core value with a low-friction one-time purchase.
Since you battled this exact Next.js issue, I’d be really curious to see what score your current setup gets today. If you run a scan, check out the new embeddable badge feature on the results page—just shipped it based on feedback from this thread!
-
3 Comments
Congrats on first dollars! Can I know how did you promote your product on this platform?
Thank you!
Honestly, I did not have a big promotion plan. The post itself was the main distribution.
What worked best here was being very specific about the problem: my own Next.js site looked fine in the browser, but the initial HTML was weak for crawlers/AI search. That made the post more like a technical lesson than a product pitch.
A few things that helped:
- I shared the actual mistake I made
- I used a concrete score/result
- I made the tool free to try without signup
- I asked people to test their own sites and share scores
- I replied to technical feedback instead of just pushing the link
The first dollars came from that discussion, not from ads or SEO. Search traffic is still basically zero for me, so I’m trying to learn how to make community distribution repeatable.
One thing I want to be clear about: this does not promise ChatGPT rankings.
The scan is about source readiness: can crawlers access, parse and summarize the public page, and are crawler policies intentional?