When someone asks an AI for a business like yours, does it name you?
More buyers now find suppliers by asking ChatGPT, Claude, Gemini or Perplexity instead of Google. These assistants recommend what they can read, and most sites are built for people, not for them: the agent arrives, understands nothing, and moves on.
We fix that. We audit your site, correct what's broken, and prove the change with a measurement you can run yourself.
- We optimise for the assistants your buyers ask:
- ChatGPT
- Claude
- Gemini
- Perplexity
The measurement is run by Ora, the independent engine behind Vercel's is-agentic.com. We don't score you ourselves: you can verify the same number independently, before and after the work.
What usually breaks (and why it costs you)
You don't need to follow the details — that's our job. But it's worth knowing what your site is losing.
| What the AI needs | What happens without it |
|---|---|
| Text present on arrivalserver-rendered HTML | Many sites only reveal content after running code. The agent doesn't wait: it sees a blank page and leaves. |
| Meaning, not just stylingsemantic structure | To you, large bold text is a heading. To a machine it's just large text. It has to be declared. |
| A briefing file for modelsllms.txt | A short file summarising who you are, what you sell and to whom. Assistants read it first. Almost no site has one yet. |
| Your facts in machine formatstructured data | Name, services, contact and location declared explicitly, instead of leaving the machine to infer them from your layout. |
| Answers ready to be quotedcitable content | If your site answers real questions in your field, the assistant can quote it and name you. If not, it names someone else. |
| Errors that don't end the visitrobots and error handling | When an agent requests a page that doesn't exist, your site should say so properly and offer where to go next. Otherwise it abandons the whole visit. |
The good news: every one of these has a concrete, bounded fix. It isn't open-ended work, which is why we can quote a fixed price before starting.
How we work
- Free scanSend us your domain and we return the score with the specific failures explained in plain language. No commitment, no call required, no company details needed.
- Fixed priceWe tell you what it costs to fix before we start — one price, no surprises. If you'd rather have your own team apply our report, that works too. No subscriptions, no lock-in.
- VerificationWe re-scan with the same public engine used for the initial scan and hand over the before and after.
Want to know where your site stands? We'll send the score and the specific failures, free and with no commitment.
The proof is in front of you
This page is the demonstration. It's built with everything we sell: served complete without JavaScript, semantic structure, structured data, llms.txt, and content in a format an assistant can quote directly.
Don't take our word for it: run maquinable.com through the public scanner and look at the score. Then run yours. That gap is exactly the work we do.
Frequently asked questions
- What is agent-readiness?
- It's how readable and usable your website is to AI agents: assistants like ChatGPT, Claude, Gemini or Perplexity, and the crawlers that feed their answers. It's measured against concrete technical criteria: content served in HTML, semantic structure, llms.txt, structured data, metadata and error handling.
- Is this just SEO?
- It's the next step. SEO optimises for Google's index; agent-readiness and GEO optimise for answer engines. They share a technical base, but agents have requirements classic SEO never covered, like llms.txt or content formatted to be quoted verbatim.
- How much does it cost?
- It depends on what needs fixing, which is why the scan is free: the scan produces one fixed price before you commit to anything. No monthly retainers, no lock-in.
- How long does a project take?
- Days, not months. Most fixes ship in one to two weeks and the verification re-scan is immediate. You don't wait months of ranking cycles to see the result.
- What exactly do you deliver?
- The files and changes implemented on your site (llms.txt, JSON-LD structured data, structure and metadata fixes, content with summaries and FAQs), plus the verification report with the before and after measurement.
- How do I know it worked?
- Because we don't do the measuring: we use a public, independent scanner you can run yourself against your own domain, before you hire us and after we deliver.
Who's behind this
Maquinable is an independent studio with years of experience implementing ERP systems and AI automation for small and mid-sized companies. Our speciality is the one you can see on this page: clean structure that machines understand on the first pass.