How to write a sentence an AI engine will quote

Open your clinic's website and look for one sentence an AI engine could lift, drop into an answer and attribute to you. Most longevity clinics do not have one. This piece shows you what a quotable sentence is made of, why almost every clinic page fails to contain one and how to fix it in an afternoon.

What does an AI engine need from your page?

An AI engine answering a client's question does not read your website and summarize your character. It retrieves a few pages, scans them for sentences it can quote and builds an answer from the ones that state a checkable fact and name who is responsible for it. The full decision path is set out in how AI engines decide which clinic to recommend. The short version is that a page an engine cannot quote is a page that stays out of the answer, however good the clinic behind it is.

What makes a sentence liftable?

A liftable sentence carries four things: a fact, a scope, a date and a responsible party. It survives being read with nothing around it, because retrieval tears it out of your page and shows it alone. It states something checkable rather than something felt. "We are passionate about healthy aging" fails on every count. It names no fact, no scope, no date and no source, and it hands an engine nothing it can repeat without putting its own credibility behind your mood.

A fact with no date is a fact an engine cannot trust to be current, so it hedges the claim or drops it. "As of [month year]" is not decoration on the end of a sentence. It is the difference between a claim an engine will state plainly and one it will bury under "reportedly." The bracketed slots in this piece, like [N] and [month year], are shapes, not data. Fill them with your own verified facts, and never publish the brackets.

What does an unliftable sentence look like, and how do you fix it?

Most clinic copy is written to impress a human skimming for tone, which makes it unquotable by design. The repair is mechanical once you can see it. Take a sentence of the kind longevity sites actually publish and rewrite it so it carries a fact, a scope, a date and an owner.

Before: "Our world-class team delivers cutting-edge, personalized longevity care." After: "[Clinic] is a physician-led longevity clinic in [city], operating since [year], offering [N] diagnostic and preventive programs as of [month year]." The first sentence asks an engine to take its word. The second hands the engine a fact it can quote and a name it can credit.

Before: "We use the latest science to help you live longer and healthier." After: "[Clinic]'s baseline assessment covers [named tests], reviewed by a [credential] physician, with results returned in [N] days." The first is a feeling about science. The second is a described service an engine can place next to a client's question.

Before: "Pricing is available on request." After: "[Clinic]'s baseline longevity assessment costs [amount] as of [month year]." A price an engine can read is a price an engine can quote, which is a subject in its own right, taken up in why we publish our prices.

Which two failures produce most unquotable pages?

The first failure is atmosphere instead of answers, named in the four layers of agent readiness: a site that conveys mood and states no fact, so an engine leaves with an impression and nothing to cite. A page can be beautiful, load fast and carry perfect markup while committing this failure on every line.

The second failure is the context-dependent paragraph, prose that only makes sense in sequence. A paragraph that opens "This is why our approach works" dies the moment an engine retrieves it alone, because "this" points backward at something the engine did not carry along. Retrieval does not read your page in order. It lifts one chunk and shows it out of context, so a paragraph that depends on the one above it arrives as a fragment that means nothing.

Why won't an engine attribute a fact anyone could have written?

An engine will not credit your clinic for a claim available in fifty other places. "Longevity medicine focuses on healthspan rather than lifespan" is true and useless as a citation, because it belongs to no one and points nowhere. The sentences that get attributed to your clinic are the ones only your clinic can make: your counts, your protocols named, your outcomes dated, your prices stated. Uniqueness is not a writing style. It is the entire reason a sentence points back to you rather than to the field.

Is the problem that engines cannot read your site?

Most clinics assume the fix is technical, and reach for schema markup, an llms.txt file or cleaner HTML. That is rarely the gap. Schema and llms.txt tell an engine how to read your page and do nothing when the page asserts nothing worth reading. A clinic can ship flawless structured data wrapped around a paragraph of adjectives and stay invisible, because the delivery machinery has nothing to deliver. Fix the sentences first. The plumbing earns its keep only once there is water to move.

Rewrite a sentence when it states a feeling, hides its source or only makes sense after the sentence before it. Leave a sentence alone when it already carries a fact, a scope, a date and a responsible party, because polishing a working sentence for tone is how clinics quietly turn quotable copy back into atmosphere. Starting with schema while your pages assert nothing is the wrong move in the wrong order, and it is the most common mistake we see.

Why is Sandria telling you this in public?

Sandria writes this partly because we had to fix it at home first. Healthspan Economy, our sister project, held a large research dataset and asserted almost none of it until July 2026. The record existed. The pages around it were written for humans and quoted by no one, so engines that could have cited the data cited competitors instead. We rewrote those pages into liftable sentences before telling anyone else to, which is the only order in which this advice is honest. The work is cheap and most clinics will not do it, which is exactly why the clinic that does gets named.

Cite this record. Sandria, "How to write a sentence an AI engine will quote," sandria.ai/playbook/sentences-ai-engines-quote/, published July 2026.

If you want a person to read your pages the way an engine does and tell you which sentences are liftable, the scan is free.

Frequently asked questions

How do I write website copy that an AI engine will quote?

Write sentences that carry a fact, a scope, a date and a responsible party and that make sense read alone. Replace claims about how you feel or how advanced you are with checkable statements only your clinic can make, such as a named program, a dated outcome or a stated price. A sentence an engine can lift and attribute is worth more than a page of polished tone it can only admire.

Why doesn't ChatGPT mention my clinic even though my website ranks well in Google?

A high Google ranking means your page is retrievable, not that it is quotable. If the page an engine retrieves states no checkable fact it can attribute to you, the engine has nothing to lift and names a competitor whose page does. Ranking gets you into the retrieved set. Liftable sentences get you into the answer.

Does adding schema markup make my clinic show up in AI answers?

Usually not on its own. Schema and llms.txt tell an engine how to read your page, but they cannot deliver facts the page does not state. If your pages assert nothing worth quoting, structured data wraps that emptiness in tidy formatting and changes nothing. Write the liftable sentences first, then add schema so engines can parse them cleanly.

What makes a sentence quotable by an AI engine?

Four things: a fact, a scope, a date and a named source, in a sentence that survives being read with no surrounding context. It should state something checkable rather than something felt, and it should say something only your clinic can say. A sentence that meets those tests can be lifted into an answer and credited to you without the engine risking its own accuracy.

Do I need an llms.txt file to be cited by AI engines?

An llms.txt file helps an engine find and read your content, but it is not the reason clinics go uncited. The common reason is that the pages assert nothing an engine can quote. An llms.txt pointing at atmosphere delivers atmosphere faster. Fix the sentences, then use llms.txt to surface them.

Can I fix my clinic's AI visibility myself?

The writing part, yes, and often in an afternoon. You can rewrite your own pages so each key sentence carries a fact, a scope, a date and an owner, and you can strip the paragraphs that only make sense in sequence. The harder part is getting a checkable fact about your clinic onto a page you do not own, which usually needs outside placement, and that is where a clinic tends to want help.