Your next customer may be an AI agent.
Can it use your website?
People now ask ChatGPT, Claude and Gemini to find a business, compare options and book for them. If an agent gets stuck on your site, that booking can quietly go to a competitor. We show — with independent proof from the real ChatGPT, Claude and Gemini agents — where each one succeeds, where it stops and what to fix first.
Start with a free fit check: we tell you whether your site fits the audit and which scenarios are worth testing — no payment, no audit run spent. Paid audits from $99/audit on the 10-pack · single audit $149 · see a sample report
Live replay of a real audit — “book an appointment”, NYC dental clinic
AI-visibility (GEO) tools measure whether assistants mention you. We test the other half: whether an AI agent can actually get through your site and finish a booking or request. Visibility in AI ≠ passability for AI — and only the second one turns AI traffic into customers.
Built for simple websites with one clear action
The audit tests one real scenario end to end — so it shines where a customer action is simple and valuable. And we're honest about where it won't help.
A great fit
A simple product or service with a clear action a customer takes on your website:
- Hotels, tours and restaurants with booking or reservation
- Dental, medical and beauty clinics with appointment requests
- Local services: HVAC, plumbing, roofing, legal, repair, cleaning
- Shops and studios with a simple catalog and an order or enquiry form
- Any site where the goal is a booking, request, call or quote
Not a fit
Where one scenario can't represent the website, the audit won't give you real value:
- Services where the main action requires signing in to a personal account (SaaS, banking, member platforms)
- Large marketplaces with many independent sellers
- Very large websites and portals of big businesses with thousands of pages
- Sites fully behind a paywall or login wall
Not sure about your website? Enter your domain above — the free fit check tells you whether the audit will help and which scenarios are worth testing. No payment, no audit run spent.
Check my website — freeOne real scenario. One honest answer.
You pick the scenario
The one action that earns you money: “book an appointment”, “request a quote”, “find pricing”, “contact sales”. You give us the URL and the scenario — that's it.
Three real AI agents attempt it
Real ChatGPT (OpenAI), Claude (Anthropic) and Gemini (Google) agents each run your scenario the way a customer's assistant would: they read pages, open menus, fill forms with clearly-marked test data. Your forms may receive up to three test submissions — one per agent, each labeled as an audit. They never pay, never solve CAPTCHAs, never fake their way past bot protection.
You get the full picture
One comparison report: Agent-Ready Score overall and per agent, a matrix of who managed what, every blocker with priority and screenshot, a video of each agent's walkthrough, and fixes written for a business owner — not just a developer.
Every audit is a complete answer
One scenario, walked end to end by three real agents. Every report is an honest snapshot of your website on the audit date — and everything below is included every time.
Agent-Ready Score
A 0–100 score for your scenario across 7 weighted categories — for each agent and overall, so you see who copes and who fails.
Fixes in plain language
Problem → why it loses you customers → how to fix it. Actionable for your developer or agency.
Blockers with priorities
Every obstacle ranked high / medium / low, with a screenshot of exactly where the agent got stuck.
Video of every agent's walkthrough
Screen recordings of ChatGPT, Claude and Gemini working through your site — see your website through each AI's eyes.
Step-by-step replay
What each agent saw, what it tried, what happened — every step of the scenario, narrated, plus a comparison matrix of who managed what.
Your next run
Blockers found? Fix and re-run the same scenario to confirm. All clear? Here's what to test next.
Ready-to-use fix files
Not just advice: a corrected robots.txt, llms.txt and structured data generated from your real site. Download, review, publish.
Before/after between runs
Re-run the same scenario after the fixes and the report opens with “What changed”: the score delta and which blockers are fixed, still there, or new.
Share link
A view-only link for your developer or agency — no sign-in needed — plus a public confirmation page with your result and the audit date.
Real audits. Real findings.
Three real clinics, same scenario — “book an appointment”. Each site was walked through by three real AI agents: ChatGPT, Claude and Gemini. Names anonymized; findings untouched.
Sample reports are real audits with business names and visuals anonymized. Your report shows your website in full detail — with each agent's video, the comparison matrix and ready-to-use fix files.
One product. Pick how many runs you need.
Every audit is the full product — report, video, score, fixes. Most sites need more than one run: fix what the first audit finds, re-run the same scenario to confirm, then test the next scenario. Not sure your site qualifies? Start with the free fit check — it costs nothing and recommends the scenarios worth testing.
- 1 audit run — one scenario, end to end
- Full report, video, score and ready-to-use fix files
- Perfect first check of your main scenario
- 3 audit runs — use them anytime, they never expire
- Audit → fix the blockers → re-run to confirm
- Or test 3 different scenarios
- 10 audit runs — cover every key scenario
- Booking, quotes, pricing, contact — verified
- Re-check after every site change
Every report ends with a clear next step: blockers found → fix and re-run the same scenario; scenario passed → which scenario to test next. Your runs are always spent on what matters most. Prices are shown in USD; accounts with Ukraine as their billing country pay in UAH (₴6,599 / ₴5,699 / ₴4,399 per audit).
Sell agent-readiness to your clients — with proof
If you build or market websites, agent-readiness is a service you can offer clients with independent proof in hand. The Scale pack — the same product, 10 audit runs at $99 each — covers a whole client portfolio: audit, hand the fixes to your team, re-run to prove the result.
- 10 audits for your clients — the existing Scale pack, runs never expire; one pack, any mix of client sites and scenarios
- Reports that travel — a view-only share link for every report, plus ready-to-use fix files your developers can publish as-is
- Before/after proof — re-run the same scenario after your fixes: the report opens with the score delta and which blockers are gone, still there or new — verified improvement you can show the client.
Reports are delivered under the UsableByAI brand (white-label is not part of the MVP). Every audit is the full product — nothing is cut for agencies.
Independent proof, not promises
Independence
The verdict doesn't depend on a single AI vendor: three different agents walk the same scenario.
Evidence
You can see what happened — video, screenshots and a step-by-step record of every walkthrough.
Transparency
Limits and failures are not hidden. If an agent is blocked or a check isn't possible, the report says so.
Repeatability
After you fix the blockers, the same scenario can be re-run to confirm the result changed.
We believe technology should make good businesses stronger — not invisible. Readiness is what we check. Trust is what we build.
Questions, answered
We don't simulate agents — every audit runs your scenario through the real computer-use agents of the three major vendors: ChatGPT (OpenAI), Claude (Anthropic) and Gemini (Google). Each reads your pages, decides what to click, fills forms and follows your site's rules. You see each agent's walkthrough on video and how their results compare.
Because one audit is a diagnosis, not a cure. The typical cycle is: audit finds blockers → you fix them → re-run the same scenario to confirm agents now get through. After that, most businesses test their other key scenarios — quotes, pricing, contact. Runs never expire, so you spend them whenever you're ready.
Yes. The agent uses clearly-marked test data (name “UsableByAI Test”, a test email), labels any submission as an automated audit, never enters payment details, never confirms a purchase and never attempts to bypass CAPTCHAs or bot protection. Depending on the scenario, your forms may receive up to three test submissions — one per agent — so you know exactly where they came from.
That's one of the most valuable findings an audit can give you. Many websites sit behind bot protection like Cloudflare, which blocks AI agents by default. If that's you, your report will show it and explain exactly how to allow verified AI agents — usually a single setting — so assistants can send you customers instead of bouncing off.
One specific action a customer would ask an assistant to do: book an appointment, request a quote, find the price of a service, submit a contact form. One audit run tests one scenario honestly, end to end.
No. We don't audit your whole site, your rankings or your design. We answer one question with evidence: can an AI agent complete your most valuable scenario? Everything in the report serves that question.
Contact us within 7 days of delivery. If you point out a concrete quality issue and we confirm it, we regenerate the report for free in the same scope. If we can't fix it — or you're simply not happy — you can request a one-time 100% refund. It applies to the purchased report; a new website or scenario is simply your next run.
Run the free fit check: enter your domain and we'll tell you whether your site meets the audit requirements and which scenarios are worth testing — order a service, find pricing, find contacts. It's free, doesn't spend an audit run, and your check history stays in your account. If a site isn't a fit — a marketplace, or a service hidden behind a login — we say so honestly before you pay.