How Thyra Health is winning AI search over legacy EHRs with Citable

Learn how Thyra Health, a full EHR for endocrinology and primary care, uses Citable to win AI search citations against legacy EHR providers in the category.

Citable's mission is to help brands win AI search by building, maintaining, and iterating a precise model of their target audience, fed by real evidence rather than keyword guesses. We strictly follow Google's E-E-A-T framework and the AI search anti-spam guidelines as our quality bar. No shortcuts, no thin content.

The results:

• #1 of 6 in Citation Rate at 19.3%, ahead of the next EHR in the category at 8.7%
• #1 of 6 in Average Rank at #1.27, the best position when mentioned
• #2 of 6 in Sentiment at +32.4, near the top of the field, with a +3.2 improving signal
• Two blog posts per week, zero hours of founder writing time

About Thyra Health

Thyra Health is a full EHR for endocrinology and primary care from Endo Mind, Inc., founded by a doctor for clinicians managing diabetes, thyroid conditions, and metabolic disorders. Rather than bolting an AI scribe onto a legacy system, Thyra unifies the scribe, the inbox, CGM interpretation, orders, and protocols around a single longitudinal patient memory, so clinicians stop juggling tools that do not talk to each other and stop spending nights and weekends on charting.

Why do legacy EHRs dominate AI search by default?

Because AI engines learned the category from decades of legacy training data. When clinicians ask which EHR offers a given workflow, AI answers default to the names with the largest historical footprint of articles, journals, product pages, and inbound links, regardless of which product is actually best for the specialty in front of them. The legacy footprint compounds inside the model. For a newer EHR built outside that stack, the visibility problem is structural, not promotional.

The legacy systems were built for billing. They were never built for clinical. Once clinical was mandated, the incumbents started pivoting and adding. If you are already making a billion a year in revenue and you have a workflow that everybody knows, it is difficult to change everything.

JJ, founder, Endo Mind, Inc. (maker of Thyra Health)

How does Citable help Thyra define its target audience?

Most AI-search tools start downstream, helping you score or distribute content you already decided to write. Citable starts upstream. For Thyra, that meant building a clinical-buyer model rooted in real evidence, not in keyword guesses. The platform pulls signals from where clinicians are actually talking about EHR workflow today: subreddits where physicians discuss inbox burden, clinical journal conversations, the AI search queries that currently surface competitors, and notes from Thyra's own sales calls. Those signals roll up into Thyra's GEO question bank, the working set of queries Thyra is qualified to answer well, scored by buyer-relevance and refreshed as the conversation shifts.

The question bank is the spec for everything downstream. Agent 1's social-listening filter scores topics against it. Agent 2's drafting engine grades content against it. As clinician concerns shift week to week, the bank updates and the content engine follows automatically.

Citable's action engine compounds on top. Every citation Thyra earns, every prompt that surfaces Thyra in an AI answer, every move in competitor performance feeds back as a signal about which content patterns are actually winning, so the next recommended action is sharper than the last. This is the layer most AI-search tools do not have: a buyer model that lives, breathes, and compounds.

How did Thyra build a content engine AI engines actually cite?

By building a two-agent workflow on top of Citable's MCP that listens for the right clinical questions and writes against a quality bar before anything ships. Both agents run on Claude, and neither agent invents anything. They surface and structure expertise JJ already brings from two decades of clinical and product work, then put every draft through Citable's GEO scoring loop until it is good enough to publish.

Agent 1 listens. Every weekday for 45 minutes, with a broader sweep on Mondays that includes the American Diabetes Association and American Medical Association, the first agent monitors where clinicians are actually discussing EHR workflow problems on Reddit, YouTube, and clinical journals. Every candidate topic gets scored against Thyra's GEO question bank, the set of queries Thyra is qualified to answer well. Topics that pass the relevance threshold land in GitHub, and JJ gets a Telegram notification to override or accept.

Agent 2 writes. Once a topic is approved, the second agent pulls the live GEO priorities, brand context, and current content performance from Citable's MCP. It drafts the post in JJ's voice, grounded in Thyra's actual product behavior. Then it scores the draft against Citable's E-E-A-T criteria and iterates until the score crosses 85. Two posts per week, automatic.

The process you have in place is so well defined that it was easy to follow. You need to get the GEO score, then it is easy to take that and publish. It was easy because it is a process. You do not have to invent anything.

JJ, founder, Endo Mind, Inc. (maker of Thyra Health)

How does Citable measure AI search impact across the funnel?

Citable scores brands on the four signals AI engines actually use to decide what to surface, in funnel order from input to quality:

• Citation Rate, how often a brand is cited as a source in an AI answer
• Mention Rate, how often a brand is named at all
• Average Rank, the brand's position when it does appear
• Sentiment, how positively the brand is described

Each layer captures a different signal AI engines weight when deciding what to cite. For Thyra, every published post hits Citable's quality threshold before it ships. The score is the gate, not a vanity metric. The agent does not publish below 85, which roughly maps to "this is content the founder would put their own name on." That standard, applied twice a week against questions clinicians actually ask, is what moved the scoreboard.

Where exactly is Thyra winning in AI search?

Across two of the four funnel layers, Thyra ranks first in the EHR category, and the other two place Thyra near the top of the field. Each layer is measured against the same five competitors.

Citation Rate, ranked first. When AI engines cite a source for EHR workflow questions, they cite Thyra's content more often than any other EHR in the category. Thyra at 19.3 percent, the next provider at 8.7 percent. Citations are the deepest signal a brand can earn in AI search: the model picked Thyra, named Thyra, and showed the link.

Mention Rate, ranked third. Thyra is named in roughly one in five EHR-related AI answers, behind only the two largest incumbents, both of which carry thousands of times Thyra's content surface. Among independent EHRs, Thyra is first by a wide margin.

Average Rank, ranked first. When Thyra is mentioned, it is mentioned first. Average position #1.27, ahead of every other EHR in the category. This is the layer that determines whether a clinician clicks through.

Sentiment, ranked second. A score of +32.4 places Thyra near the top of the field, with a +3.2 improvement, the strongest improving signal on the scoreboard. AI engines describe Thyra more positively because the underlying content carries the voice of someone who has actually used an EHR in clinical practice.

What does this free a founder up to do?

Before Citable, most founders either do not publish at all or publish into a vacuum. They sit down at a blank page, guess at what their buyer cares about, draft something, second-guess it, rewrite, and never quite know whether the result is going to win an AI citation. The bottleneck is cognitive, not operational. With Citable, JJ knows exactly who Thyra is writing for, which questions those clinicians are asking AI engines today, and what bar each draft has to clear before it ships. He approves the topic, the agents handle the drafting and iteration against the score, and the work left for JJ is the work that compounds: talking to doctors, running demos, shaping the product.

I can actually talk to doctors, do the demos, and talk about onboarding.

JJ, founder, Endo Mind, Inc. (maker of Thyra Health)

A clinician who had been testing different EHRs for six months reached out to sign her group up after a single demo. The work only a founder can do is the work JJ is now spending his time on.