Doximity, Inc. (DOCS) 2027 Q1 法說會逐字稿

Ladies and gentlemen, thank you for standing by. My name is Abby, and I will be your conference operator today. At this time, I would like to welcome everyone to the Doximity first-quarter 2027 earnings conference call. (Operator Instructions)

各位女士、先生，感謝您稍候。我叫 Abby，今天將擔任本次電話會議的接線員。此刻，我謹代表公司歡迎各位參加 Doximity 2027 年第一季財報電話會議。（接線員指示）

And I would now like to turn the conference over to Perry Gold, Senior Vice President, Investor Relations. You may begin.

接下來我想把電話會議交給投資人關係資深副總裁 Perry Gold。您可以開始了。

Thank you, operator. Hello, and welcome to Doximity's fiscal 2027 first-quarter earnings call. With me on the call today are Jeff Tangney, Co-Founder and CEO of Doximity; and Matt Sonefeldt, CFO.

謝謝，接線員。各位好，歡迎參加 Doximity 2027 會計年度第一季財報電話會議。今天與我一同出席的有 Doximity 共同創辦人兼執行長 Jeff Tangney，以及財務長 Matt Sonefeldt。

A complete disclosure of our results can be found in our press release issued earlier today as well as in our related Form 8-K along with a copy of our prepared remarks, all available on our website at investors.doximity.com.

我們的完整業績揭露可見於今日稍早發布的新聞稿，以及相關的 Form 8-K；其中亦包含我們事先準備的講稿副本，皆可於我們網站 investors.doximity.com 取得。

As a reminder, today's call is being recorded, and a replay will be available on our website. As part of our comments today, we will be making forward-looking statements. These statements are based on management's current views, expectations and assumptions and are subject to various risks and uncertainties. Actual results may differ materially and we disclaim any obligation to update any forward-looking statements or outlook.

提醒各位，今天的電話會議將被錄音，並會在我們網站提供重播。在今天的發言中，我們將作出前瞻性陳述。這些陳述係基於管理層目前的觀點、預期與假設，並受各種風險與不確定性影響。實際結果可能有重大差異；我們亦不承擔更新任何前瞻性陳述或展望的義務。

Please refer to the risk factors in our annual report on Form 10-K and any subsequent Form 10-Qs and other reports and filings with the SEC that may be filed from time to time, including our upcoming filing on Form 10-Q.

請參閱我們 Form 10-K 年報中的風險因素，以及其後不時向美國證券交易委員會（SEC）提交的 Form 10-Q 與其他報告及申報文件，包括我們即將提交的 Form 10-Q。

Our forward-looking statements are based on assumptions that we believe to be reasonable as of today's date, August 6, 2026. Of note, it is Doximity's policy to neither reiterate nor adjust the financial guidance provided on today's call unless it is also done through a public disclosure such as a press release or through the filing of a Form 8-K.

我們的前瞻性陳述係基於截至今日（2026 年 8 月 6 日）我們認為合理的假設。特別說明，Doximity 的政策是：除非也透過新聞稿或提交 Form 8-K 等公開揭露方式進行，否則不會在今日電話會議中重申或調整所提供的財務指引。

Today, we will discuss certain non-GAAP metrics that we believe aid in the understanding of our financial results. A historical reconciliation to comparable GAAP metrics can be found in today's earnings release. Finally, during the call, we may offer incremental metrics to provide greater insights into the dynamics of our business. These details may be onetime in nature, and we may or may not provide updates on those metrics in the future.

今天我們將討論若干非 GAAP 指標，我們相信這些指標有助於理解我們的財務表現。與可比 GAAP 指標的歷史調節表可見於今日的財報新聞稿。最後，在會議期間，我們可能會提供額外指標，以更深入說明我們業務的動態。這些細節可能屬一次性資訊，未來我們可能會或可能不會再更新這些指標。

I would now like to turn the call over to our CEO and Co-Founder, Jeff Tangney. Jeff?

接下來我想把電話交給我們的執行長兼共同創辦人 Jeff Tangney。Jeff？

Thanks, Perry, and thanks, everyone, for joining our first-quarter earnings call. Today, I'll cover four things: our financials, AI study results, usage growth and commercial AI progress.

謝謝你，Perry，也謝謝各位參加我們第一季財報電話會議。今天我將涵蓋四個主題：我們的財務表現、AI 研究結果、使用量成長，以及商業化 AI 的進展。

First, our financials. Revenue growth reaccelerated to $157 million in Q1, up 7% year on year. Adjusted EBITDA was $75 million or a margin of 48%. Respectively, these were bold beats of 3% and 8% versus the high end of our guidance. As you'll hear again shortly from our CFO, Matt, this is our AI investment year. We're proving you can still post best-in-class software margins while investing heavily in clinical AI. We're leaning in as we see a once-in-a-generation opportunity to build the new AI age of medicine.

首先是財務表現。第一季營收成長重新加速至 1.57 億美元，年增 7%。調整後 EBITDA 為 7,500 萬美元，或 48% 的利潤率。相較於我們指引區間上緣，這兩項分別大幅超出 3% 與 8%。正如稍後你們也會再次聽到我們財務長 Matt 提到的，今年是我們投資 AI 的一年。我們正在證明：即使大力投資臨床 AI，仍然可以維持同級最佳的軟體利潤率。我們正加大投入，因為我們看到這是一個世代難得的機會，去打造醫療領域全新的 AI 時代。

To that end, I'm proud to report that we're seeing record AI usage while topping the first large-scale independent head-to-head trial of clinical AI vendors. A few weeks ago, a team of 57 researchers led by 29 physicians from Stanford and Harvard published NOHARM, the first-ever independent study of 24 clinical AI models and how they perform in 1,100 real-world patient cases. It's the kind of rigorous independent physician-led research that we need more of.

基於此，我很自豪地報告：我們正看到創紀錄的 AI 使用量，同時在首個大規模、獨立的臨床 AI 供應商正面對決試驗中名列第一。幾週前，由 57 位研究人員組成、其中包含來自史丹佛與哈佛的 29 位醫師所領導的團隊，發表了 NOHARM——這是史上第一個針對 24 個臨床 AI 模型的獨立研究，評估其在 1,100 個真實世界病患案例中的表現。這正是我們需要更多的、嚴謹且由醫師主導的獨立研究。

Our Doximity ask product led among US models with the lowest clinical error rates and the highest safety ratings or as Fortune Magazine, Doximity asked, came out on top. Clinical AI is rapidly improving. Our winning model demonstrated a 4.8% air rate, while others like Anthropic's Best model, Fable 5 finished with a 13.6% error rate.

我們的 Doximity Ask 產品在美國模型中領先，具備最低的臨床錯誤率與最高的安全評等；正如《財富》雜誌所報導的，Doximity Ask 脫穎而出、名列第一。臨床 AI 正在快速進步。我們的勝出模型錯誤率為 4.8%，而其他模型例如 Anthropic 的最佳模型 Fable 5，最終錯誤率為 13.6%。

We believe our outperformance is due mainly to two things: one, our unique built-in drug reference a model within the model, which is 100% expert verified to ensure accurate drug doses and interactions; and two, our over 12,000 physician per check editors who are continuously reviewing and refining our AI outputs. These safeguards and quality checks are critical for hospital AI steering committees who could be held liable for their outputs and therefore, care deeply about their accuracy.

我們認為我們的優異表現主要來自兩點：第一，我們獨特的內建藥物參考——也就是「模型中的模型」，並經 100% 專家驗證，以確保藥物劑量與交互作用的準確性；第二，我們超過 12,000 名醫師的逐次審核編輯團隊，持續審閱並精煉我們的 AI 輸出。這些防護機制與品質檢核，對於醫院的 AI 指導委員會至關重要；因為他們可能需對輸出結果承擔責任，因此非常重視其準確性。

We continue to lead the way in the enterprise now with 165 signed health system AI clients, including 8 of the nation's top on a roll hospitals. Our recent wins include Northwestern, Penn Medicine and the University of Michigan. As this market migrates from AI Wild West to privacy and risk management, we're well positioned to win as we did in telehealth.

我們在企業端持續領先，目前已簽約 165 家醫療體系 AI 客戶，其中包括全美排名前列的 8 家「榮譽榜」醫院。我們近期的勝利案包括 Northwestern、Penn Medicine 以及密西根大學。隨著市場從 AI 的「蠻荒西部」走向重視隱私與風險管理，我們具備良好條件再次勝出，就像我們在遠距醫療領域所做到的一樣。

Okay. Now to our usage growth. Quarterly active workflow prescribers grew more than 30% year on year to record highs with nearly half using our AI tools in Q1. I AI prompt volume was up more than 25% quarter on quarter, while our AI scribe notetaking users we were whopping 10 times this July over prior. With these gains, we believe we're now the only clinical AI company who is top three in both the AI search and scribe markets. Finally, our commercial AI products.

好。接下來談使用量成長。季度活躍工作流程開立處方醫師（workflow prescribers）年增超過 30%，創下新高，其中近一半在第一季使用了我們的 AI 工具。AI 提示（prompt）量較上一季成長超過 25%，而我們的 AI 書記（scribe）筆記使用者數在今年 7 月較先前暴增至 10 倍。憑藉這些成長，我們相信我們現在是唯一一家在 AI 搜尋與 AI 書記兩個市場都名列前三的臨床 AI 公司。最後，談談我們的商業化 AI 產品。

Our AI search monetization is off to a strong start, and the higher-level conversations it's generating with clients are fueling new business across our broader pharma portfolio. New search contracts are driving our revenue raise for the year, and we're just getting started. 15 years ago, we carved out a niche as the leading online resume book for physicians.

我們的 AI 搜尋變現進展強勁，而它所帶來的、與客戶更高層級的對話，正在推動我們更廣泛的製藥產品組合新增業務。新的搜尋合約正在帶動我們上調本年度營收預期，而我們才剛剛開始。15 年前，我們以醫師線上履歷資料庫的領導者身分切入市場並建立利基。

We've grown a lot since then by keeping clinicians first and adapting the latest tech to their needs. Today, we believe we're the number one most used clinical service in at least 5 categories: networking, news, scheduling, facts and telehealth. We're the doctor's digital platform and AI is just the next chapter in our growth.

此後，我們始終以臨床人員為優先，並將最新科技調整以符合他們的需求，因此成長良多。今天，我們相信我們至少在 5 個類別中是使用率最高的臨床服務第一名：人脈連結、新聞、排程、資料查詢與遠距醫療。我們是醫師的數位平台，而 AI 只是我們成長的下一個篇章。

As always, I'd like to end by thanking my Doximity teammates who continue to work incredibly hard the care for those who care for us.

一如既往，我想在最後感謝 Doximity 的團隊夥伴們，他們持續非常努力地工作——照顧那些照顧我們的人。

With that, I'll hand it over to our CFO, Matt Sonefeldt, to walk through our financials and guidance. Matt?

接下來我把時間交給我們的財務長 Matt Sonefeldt，請他帶大家回顧我們的財務表現與指引。Matt？

Thanks, Jeff.

謝謝你，Jeff。

Q1 '27 was a strong quarter for Doximity with robust revenue growth in clinicians adopting our AI suite faster than anticipated. Higher-than-expected AI usage creates a good problem for Doximity, and will expand our AI investment in fiscal '27 to capture the significant long-term opportunity ahead.

對 Doximity 而言，2027 年第一季表現強勁：營收成長穩健，且臨床人員採用我們的 AI 套件速度快於預期。高於預期的 AI 使用量對 Doximity 來說是個「甜蜜的煩惱」，也將使我們在 2027 會計年度擴大 AI 投資，以掌握前方重大的長期機會。

Turning to our top line. Q1 '27 revenue of $157 million outperformed the high end of our guidance. with growth improving to 7% year over year. Revenue growth rebounded with solid performance across both pharma and hospital customers. Continued strong demand from large customers highlights our growing opportunity to work with pharma and hospital CXOs.

接著看營收表現。2027 年第一季營收為 1.57 億美元，優於我們指引區間上緣，年增率提升至 7%。營收成長回升，製藥與醫院客戶兩端皆有穩健表現。大型客戶持續強勁的需求，凸顯我們與製藥公司及醫院 CXO（高階主管）合作的機會正在擴大。

Our largest customers continue to drive our growth. We now have 127 pharma and hospital customers to generate more than $500,000 in annual subscription revenue on a trailing 12-month basis, representing 7% growth year over year. They contributed 83% of total revenue, a level consistent with prior quarters. The top 20 customers produced net revenue retention, or NRR, of 112% with overall NRR and 107% in Q1 on a trailing 12-month basis.

我們最大的客戶持續推動我們的成長。以過去12個月（TTM）計算，我們目前有127家製藥與醫院客戶可帶來超過50萬美元的年度訂閱收入，年增7%。他們貢獻了總營收的83%，與前幾季的水準一致。前20大客戶的淨營收留存率（NRR）為112%，而整體NRR在Q1以過去12個月計算為107%。

Q1 outperformance was driven by two factors specific to pharma customers. First, our new AI search product drove higher overall client engagement. As a reminder, we launched AI search in late April, leading to an increased velocity of pharma customer interactions. These conversations supported overall demand even as we did not recognize any AI revenue in Q1.

Q1的優於預期表現主要由兩項製藥客戶特有的因素所帶動。首先，我們新的AI搜尋產品帶動整體客戶互動度提升。提醒一下，我們在4月下旬推出AI搜尋，使製藥客戶互動的速度加快。即使我們在Q1尚未認列任何AI營收，這些對話仍支撐了整體需求。

We have onboarded our first cohort of AI search customers across more than two dozen programs. We're also building a healthy pipeline for the remainder of FY27 and beyond with robust demand for our trusted brand and rigorously verified NPI level engagement. We expect the majority of AI search revenue contracted to date to be recognized during Q3.

我們已在超過二十多個專案中，完成首批AI搜尋客戶的導入。同時，憑藉我們值得信賴的品牌與經嚴格驗證、達到NPI（國家提供者識別碼）層級的互動，我們也為FY27剩餘期間及之後建立了健康的商機管線，需求強勁。我們預期目前已簽約的AI搜尋營收多數將於Q3認列。

The second factor was unlocking additional budget from several customers that only committed to shorter-term buys during last year's upfront. For example, we saw a meaningful rebound in spend from one of our large top 20 pharma customers that spent less than Q3 of last year. While the overall pharma spending environment remains tight, we're starting to win innovation budget with the launch of AI search. Turning to profitability.

第二項因素是，我們從數位客戶那裡解鎖了額外預算；這些客戶在去年預售（upfront）時僅承諾較短期的採購。例如，我們看到某家前20大製藥大客戶的支出明顯回升；其支出曾低於去年Q3的水準。雖然整體製藥支出環境仍然吃緊，但隨著AI搜尋的推出，我們開始贏得創新預算。接下來談獲利能力。

Adjusted EBITDA in Q1 '27 was $75 million, representing a 48% margin. The flow-through of incremental revenue growth drove the Q1 outperformance versus our outlook. In Q1, non-GAAP gross margin was 88% versus 91% last year. We increased AI compute spend during the quarter to support higher-than-expected clinician AI usage. We expect to maintain this trend throughout fiscal '27 as we doubled down on AI investments to further scale ask engagement.

27會計年度Q1的調整後EBITDA為7,500萬美元，代表48%的利潤率。增量營收成長的流入（flow-through）帶動Q1表現優於我們的展望。Q1的非GAAP毛利率為88%，去年同期為91%。本季我們提高了AI運算支出，以支援高於預期的臨床醫師AI使用量。我們預期在27會計年度全年維持此趨勢，因為我們加碼AI投資以進一步擴大ask互動規模。

During Q1, we also saw higher costs in our other OpEx lines driven by annual merit increases, greater internal AI usage and brand marketing. On a GAAP basis, stock-based compensation, or SBC, was $37 million in the quarter or 23% of revenue. This is consistent with the low 20% SBC guidance provided last quarter.

在Q1期間，我們也看到其他營業費用（OpEx）項目成本較高，主要由年度績效調薪、更多內部AI使用以及品牌行銷所帶動。以GAAP基礎計算，本季股票基礎薪酬（SBC）為3,700萬美元，占營收23%。這與上季提供的SBC指引（約低20%區間）一致。

As a reminder, this year's SBC increase is primarily related to the fiscal '26 grant made to our AI-focused R&D team. Excluding the grant, SBC would have been approximately 19% of revenue in Q1. We also saw a small impact from the hiring of new executives.

提醒一下，今年SBC增加主要與26會計年度授予我們以AI為重點的研發團隊之獎勵有關。若排除該次授予，Q1的SBC約為營收的19%。我們也因聘用新任高階主管而受到些微影響。

Our GAAP effective tax rate was approximately 40% in the first quarter compared to 17% in the prior year, driven by the tax treatment of equity compensation. Our non-GAAP effective tax rate remained at 21%. GAAP EPS was $0.13 per share and non-GAAP EPS was $0.29 per share in Q1. Fully diluted shares declined by 10 million year over year or 5% to 191 million shares outstanding.

第一季GAAP有效稅率約為40%，相較去年同期的17%上升，主要受股權薪酬稅務處理方式影響。我們的非GAAP有效稅率維持在21%。Q1的GAAP每股盈餘（EPS）為0.13美元，非GAAP EPS為0.29美元。完全稀釋股數年減1,000萬股（或5%），至1.91億股流通在外。

Our balance sheet and cash flow generation remains strong and create a solid foundation for our growth and AI investment. We ended Q1 '27 with $688 million in cash, cash equivalents and marketable securities and we remain debt-free. In Q1, we generated free cash flow of $40 million. The decrease versus the prior year was driven by normal fluctuations in collections, which have variability based on program and delivery timing.

我們的資產負債表與現金流創造能力仍然強勁，為成長與AI投資奠定穩固基礎。27會計年度Q1結束時，我們持有6.88億美元的現金、約當現金與有價證券，且仍維持零負債。Q1我們產生4,000萬美元的自由現金流。相較去年同期的下降主要來自收款的正常波動；收款會因專案與交付時點而有所變動。

We expect collections to normalize throughout the remainder of the year.

我們預期在今年剩餘期間，收款將回歸正常水準。

During the first quarter, we repurchased $92 million worth of shares. We believe share repurchases remain an attractive opportunistic use of capital. As of June 30, we had approximately $400 million remaining in our existing repurchase program.

第一季期間，我們回購了9,200萬美元的股票。我們認為股票回購仍是具吸引力、可伺機運用的資本配置方式。截至6月30日，我們現有回購計畫尚餘約4億美元額度。

Let's turn to our outlook. For Q2 '27 revenue, we expect a range of $170 million to $171 million, representing a midpoint of 1% year over year. For the full year, we have revised our guidance range up by $6 million between $671 million and $681 million, representing 5% growth at the midpoint. This increase represents the flow-through from Q1 outperformance plus a modest incremental raise.

接下來談展望。27會計年度Q2營收我們預期介於1.70億至1.71億美元，中位數代表年增1%。全年方面，我們將指引區間上調600萬美元至6.71億至6.81億美元，中位數代表5%成長。此上調反映Q1優於預期的流入效應，並加上小幅的額外上修。

The stronger fiscal '27 outlook reflects a more stable pharma budget environment, a higher velocity of customer interactions and the nascent growing AI commercial pipeline. Our Q2 '27 growth outlook is impacted by the tough comparison against last year's elevated 23% growth. In addition, we expect only modest revenue from AI search in Q2.

27會計年度較強的展望反映製藥預算環境更趨穩定、客戶互動速度提升，以及初期成長中的AI商業化管線。我們對27會計年度Q2的成長展望受到去年同期23%高成長的艱難比較基期影響。此外，我們預期Q2來自AI搜尋的營收僅為小幅。

In Q3 '27, we expect stronger year-over-year growth as AI search revenue builds, we have a more normal growth comparison to the prior year. To date, we've taken a deliberate approach to scaling AI search by protecting the ask user experience while iterating on the product based on customer feedback.

在27會計年度Q3，隨著AI搜尋營收逐步建立，加上與去年同期相比的成長比較更為正常，我們預期年增成長將更強勁。截至目前，我們在擴大AI搜尋規模上採取審慎策略：在保護ask使用者體驗的同時，依據客戶回饋持續迭代產品。

At launch, AI search programs had conservative inventory caps with shorter three- to four-month commitments. As we move into the upfront, the focus shifts towards larger, longer customer contracts with greater inventory available across more therapeutic categories.

在推出初期，AI搜尋專案採取保守的庫存上限（inventory caps），並以較短的三到四個月承諾期為主。隨著我們進入預售（upfront）期，重點將轉向更大、期限更長的客戶合約，並在更多治療領域提供更充足的庫存。

For adjusted EBITDA in Q2 '27, we expect a range of $80.5 million to $81.5 million, representing a 48% adjusted EBITDA margin at the midpoint. For fiscal '27, we now expect a revised range of $309 million to $329 million, representing a 47% adjusted EBITDA margin at the midpoint. This continues to be our AI investment year as we respond to stronger-than-expected clinician usage and our growing commercial AI pipeline. This is the right long-term decision for our members in our business.

27會計年度Q2的調整後EBITDA，我們預期介於8,050萬至8,150萬美元，中位數代表48%的調整後EBITDA利潤率。27會計年度全年方面，我們目前預期修正後區間為3.09億至3.29億美元，中位數代表47%的調整後EBITDA利潤率。今年仍是我們的AI投資年度，以因應高於預期的臨床醫師使用量以及持續成長的商業AI管線。這是對我們會員與我們業務而言正確的長期決策。

This additional spend allows us to further invest in our AI competitive advantages, increased safety and accuracy through Pure Check AI integration across our platform and our growing presence in US hospitals. Approximately 90% of AI expenses will focus on responding to increased demand for our clinical AI suite, this will be recognized in cost of revenue, and we expect gross margins to trend in the mid- to high 80% range throughout the year.

這些額外支出使我們能進一步投資於AI競爭優勢，透過在全平台整合Pure Check AI來提升安全性與準確性，並擴大我們在美國醫院的布局。約90%的AI費用將聚焦於回應臨床AI產品套件需求增加；這些費用將認列於營收成本（cost of revenue），我們預期全年毛利率將維持在80%中段至高段區間。

In summary, we're leaning in from a position of strength during one of the most important technology shifts in medical history. We will leverage our best-in-class margins and main culture to continue delivering leading AI experience for US clinicians. We believe this puts us one big step closer to realizing our mission: to help doctors be more productive so they can provide better care for their patients.

總結來說，在醫療史上最重要的技術轉變之一中，我們正以強勢基礎加大投入。我們將運用同級最佳的利潤率與主要文化，持續為美國臨床醫師提供領先的AI體驗。我們相信這讓我們更接近實現使命：協助醫師提升生產力，從而為病患提供更好的照護。

Lastly, this quarter, we've begun to include a modeling considerations appendix, where you can see more detailed financial commentary.

最後，本季我們開始加入「建模考量」附錄，您可在其中看到更詳細的財務評論。

With that, I will turn it over to the operator for questions.

接下來我將交給接線員進行問答。

(Operator Instructions) Brian Peterson, Raymond James.

(接線員指示) Brian Peterson，Raymond James。

Jeff, I wanted to start on the NOHARM study that we released a few weeks ago. I know you mentioned that in the prepared remarks. What do you think that means from a competitive perspective? And how do you ultimately think that you can influence trust and usage for physicians on the platform?

Jeff，我想先從我們幾週前發布的 NOHARM 研究談起。我知道你在事先準備的發言中有提到。從競爭的角度來看，你覺得這代表什麼意義？以及你最終認為，如何在平台上影響醫師的信任與使用？

Brian, thanks for the question. So first, I'll just say, we're really pleased with the team's efforts. It's really a testament to the hard work of our 800-person team here that they've really leaned into our pathway acquisition of almost exactly one year ago. built a unique built-in drug reference, which is really key to answering clinical questions well and then brought on board 12,000 sited authors, physician check editors who are continuously reviewing and improving our AI outputs.

Brian，謝謝你的提問。首先我想說，我們對團隊的努力感到非常滿意。這確實證明了我們這裡 800 人團隊的辛勤付出：他們確實全力投入我們大約一年前的 Pathway 收購，打造出一個獨特的內建藥物參考資料庫，這對於把臨床問題回答好非常關鍵，並且延攬了 12,000 名具引用來源的作者與醫師審稿編輯，持續審查並改進我們的 AI 輸出。

There's actually been a couple of studies that have been published now. One was done by NYU and published in nature and then the Stamford Harbor study, the NOHARM study. really groundbreaking studies that are important for clinical AI because at the end of the day, we're treating lots and lots of people, hundreds of millions of people with AI these days, and we want to make sure that they're getting the most accurate, safest answers.

其實現在已經有幾項研究發表了。其中一項由 NYU 完成並發表在《Nature》，另外還有 Stamford Harbor 的研究，也就是 NOHARM 研究。

And so the no-harm study was the first all model comparison done sort of on a secret shopper basis where they went out and asked questions without the models knowing, truly independent 1,100 real-world patient cases that they had seen at Stanford and then they went and the graded each of those answers in a consistent format.

因此，no-harm 研究是第一個「全模型比較」的研究，採用類似神秘客的方式：研究者在模型不知情的情況下提出問題；使用的是 1,100 個真正獨立的真實世界病患案例（他們在 Stanford 看過的案例），然後以一致的格式對每一個答案進行評分。

And again, we're very pleased to have come out on top of that overall approach. And I'll just say, it's the kind of rigorous independent physician-led research that we need much more of. And I hope that folks keep doing it because competition is good for the patient. It's how we all get better. And we are certainly in favor of honest competition here to have the best results, the best answers.

再次強調，我們對於在整體表現上名列前茅感到非常高興。我也想說，這正是我們需要更多的那種嚴謹、獨立、由醫師主導的研究。我希望大家能持續做下去，因為競爭對病患是好事。這是讓我們所有人都變得更好的方式。而我們當然支持誠實的競爭，以取得最佳結果、最佳答案。

I'll say we're -- I think this really plays out in the marketplace is with hospitals. Every hospital has an AI steering committee these days. And they are very much in tune with what's a high-quality output and what is not because they are ultimately liable for some of these outputs. So we're really pleased we continue to lead the way in the enterprise with 165 signed health system clients.

我想說——我認為這在市場上的呈現，主要是在醫院端。現在每家醫院幾乎都有 AI 指導委員會。他們非常清楚什麼是高品質輸出、什麼不是，因為他們最終要為其中一些輸出負責。因此，我們很高興能持續在企業端領先，目前已簽約的醫療體系客戶達 165 家。

I would liken this maybe to what's happened in the frontier or voting assistant AI space with Cloud and Anthropic over the last year. We saw a shift somewhere early this year where it wasn't just individual decision to go to whatever I want to use. It was -- it became an enterprise decision and we think the same thing will happen here in the health care space as more of these studies start to point the enterprise to the right direction.

我會把這比作過去一年在前沿（frontier）或投票助理 AI 領域中，Cloud 和 Anthropic 所發生的情況。我們在今年稍早看到一個轉變：不再只是個人決定要用什麼。而是——它變成了一個企業層級的決策；我們認為隨著更多研究開始把企業導向正確方向，醫療照護領域也會發生同樣的事。

But also as these enterprises get, I think, rightfully concerned about the leakage of patient data and what's called PHI, protected health information out to the broader Internet I can share that we see -- it's almost 30% of the questions or prompts include some form of PHI or patient identifiable information. And again, as a health system or a hospital, you don't want that spraying out across eight different websites.

此外，當這些企業開始——我認為也是理所當然地——擔心病患資料外洩，以及所謂的 PHI（受保護的健康資訊）流向更廣泛的網際網路時，我可以分享的是：我們看到——將近 30% 的問題或提示詞都包含某種形式的 PHI 或可識別病患身分的資訊。同樣地，作為醫療體系或醫院，你不希望這些資訊被散播到八個不同的網站上。

You need to make sure that's with someone that you have a privacy agreement in place with and as the privacy agreement that is closely followed. So we believe that this will shift the market to more of an enterprise motion. And again, we're proud to be leading the way there with our recent wins at Northwestern, 10 and Michigan, we look forward to leading into this market like we did in the telehealth market.

你必須確保這些資料是交給你已簽有隱私協議的合作方，而且該隱私協議會被嚴格遵循。因此我們相信，市場將轉向更偏企業端的推進模式。同時，我們也很自豪近期在 Northwestern、10 以及 Michigan 的勝利讓我們持續領先；我們期待像當年在遠距醫療市場一樣，持續帶領這個市場。

And over the course of the next couple of years, really winning as the best, most accurate clinical AI, but also the safest, most private clinical AI because both will be needed for doctors and patients.

而在接下來幾年裡，真正以最佳、最準確的臨床 AI 取勝，同時也成為最安全、最重視隱私的臨床 AI，因為這兩者對醫師與病患都同樣必要。

I appreciate all the color. And I noticed you guys did raise the outlook by a bit more than the beat this quarter. I'm curious on what you're thinking about the overall budget growth for calendar year '26. I know you mentioned maybe some innovation budgets are opening up to you, but how would you frame the overall buying environment.

感謝你提供這麼多細節。我也注意到你們把財測上調的幅度，比本季超預期的幅度還要更大一些。我想了解你們對 2026 曆年整體預算成長的看法。我知道你提到可能有一些創新預算正在對你們開放，但你會如何描述整體採購環境？

Brian, this is Matt. I think the overall buying environment still feels tight, but certainly more stable. And I think for us, especially as we come into market, was the AI search products, we're increasingly playing now in the AI innovation budget. We're playing now in kind of more of an insight and analytics budget and we're playing -- starting to play. It's early days, but in the search market overall as well.

Brian，我是 Matt。我認為整體採購環境仍然偏緊，但確實更穩定了。而對我們來說，特別是當我們以 AI 搜尋產品進入市場時，我們現在愈來愈多是在 AI 創新預算中競爭。我們也在更偏洞察與分析（insight and analytics）的預算中競爭，並且——開始涉足（雖然還在早期）整體搜尋市場。

It opens up a lot of opportunity for us. I'll manage your expectations and say it's still pretty early in the ramp. But like we've talked about in the prepared remarks, we have a couple of dozen preliminary customers under contract, and we're building out a pretty strong pipeline. So I think that helps us have a little bit more confidence in the market outlook overall.

這為我們帶來很多機會。我先管理一下你的預期：目前在爬坡期仍算相當早期。但如同我們在事先準備的發言中提到的，我們已有數十家初期客戶簽約，並且正在建立相當強的銷售管線。因此我認為，這讓我們對整體市場展望更有一些信心。

Craig Hettenbach, Morgan Stanley.

Craig Hettenbach，Morgan Stanley。

Just following up on kind of the year of AI investment as you guys lean in here, and you've talked previously in terms of a little bit of a mismatch where you're investing in the AI search will come I'm not asking for fiscal '28 guidance, but how do you think about just kind of the spend that you have now?

我想追問一下，關於你們在這裡加大投入、把今年視為 AI 投資之年這件事；你們之前談過某種程度上的不匹配：你們在 AI 搜尋上的投資會在——我不是在問 2028 財年的指引，但你們如何看待目前的支出水準？

And then ultimately, the operating leverage that you might see down the line as AI search ramps up?

以及最終，當 AI 搜尋開始放量時，你們可能在後續看到的營運槓桿？

Craig, this is Jeff. I'll take a crack at that. So first, I'll highlight what was in Matt's prepared remarks, which is 90% -- over 90% of our AI spend is in the service of doctors, right, helping them get the very best answer. And so it's not a matter of internally token maxing or doing lots of internal tools. It really is serving the broader medical marketplace and again, seeing better-than-expected volume and usage there. In terms of the economics of the usage, it's early days on our AI search product.

Craig，我是 Jeff。我來試著回答。首先我想強調 Matt 在事先準備的發言中提到的：我們 90%——超過 90% 的 AI 支出，都是為了服務醫師，對吧，幫助他們得到最好的答案。因此這不是在內部把 token 用到極致，或做很多內部工具。它真正是在服務更廣泛的醫療市場，而且我們也看到那裡的量與使用率優於預期。至於使用的經濟性，我們的 AI 搜尋產品仍在早期階段。

But I can tell you, we're earning more than 10x per search in revenue than it cost us to run that today. And again, over time, we probably expect the overall AI cost, if anything, go down as models get more efficient. So we feel good about the unit economics that are being in line with our broader gross margins or maybe even slightly better.

但我可以告訴你，我們目前每一次搜尋帶來的營收，是我們執行該搜尋成本的 10 倍以上。而且隨著時間推進，我們可能預期整體 AI 成本若有變化，反而會下降，因為模型會變得更有效率。因此我們對單位經濟性很有信心，會與我們整體毛利率水準一致，甚至可能略好一些。

And frankly, the TAM that this unlocks for us within pharma has been a real surprise and upside for us. I mean, I'll make the joke I've been at this for 15 years here at Doximity. I'm getting into conference rooms. I never got into before up on high-level floors with C-suite officers because this really is a C-level decision and C-level opportunity and focus for our pharma clients because AI is so important to how the future of AI will work and how medical decisions will fundamentally be made.

坦白說，這為我們在製藥領域解鎖的 TAM（總可服務市場）對我們而言是個很大的驚喜與上行空間。我開個玩笑：我在 Doximity 做了 15 年。我現在能走進一些會議室——以前從來進不去的那種，在高樓層、與 C-suite 高層主管會面；因為這確實是一個 C 級決策、也是 C 級機會，對我們的製藥客戶而言是 C 級的關注焦點，因為 AI 對未來 AI 將如何運作、以及醫療決策將如何從根本上被做出，實在太重要了。

So we're excited to have those discussions, and it's really up-leveling our business across the board. So in general, we're excited about the margin opportunity that AI search has for us in fiscal '28 and beyond.

因此，我們很期待進行這些討論，而這確實正在全面提升我們的業務水準。總體而言，我們對 AI 搜尋在 2028 財年及之後為我們帶來的毛利機會感到振奮。

That's helpful. And then just a follow-up from that. Understanding it's a very short period of time you're in the seat. But what has surprised you most compared to the view you had externally before you joined? And I say that, particularly at a time where we're talking a lot about technology and its AI transition.

這很有幫助。接著就此再追問一下。我理解你上任的時間非常短。但相較於你加入前從外部看到的觀點，最讓你意外的是什麼？我之所以這麼問，特別是因為我們正處在一個大家大量討論科技及其 AI 轉型的時點。

So there's a lot of things in the marketplace that people are trying to digest and anything you would call out in terms of -- again, surprised you the most about how Doximity is positioned and where you guys are going?

市場上有很多事情大家都在試著消化；就你而言，有沒有什麼可以點出的——再一次——最讓你驚訝的地方，關於 Doximity 的定位以及你們將走向何方？

Yes. Great question, Craig. I think 1 of the things that I have remarked on, especially from some of my consumer web days, especially at LinkedIn, as just how strong our relationship is with our users and members. The engagement that we have with each member on Doximity would have made LinkedIn blush.

是的。Craig，問得很好。我認為我特別有感的一點——尤其是我過去做消費者網路的經驗、特別是在 LinkedIn 的時候——就是我們與使用者與會員之間的關係有多麼強。我們在 Doximity 與每位會員之間的互動黏著度，會讓 LinkedIn 都自嘆不如。

And I think what's really exciting is to see just the rapid increase, both in terms of the number of users that are starting to use the clinical search tool as well as how they're using it, right? I think we talked about pumps increasing by 25% quarter over quarter. That's a big engagement change from the base of formations that are already highly engaged.

而我覺得真正令人興奮的是，看到臨床搜尋工具的使用者數量快速增加，以及他們使用方式的快速演進，對吧？我想我們提到過，查詢量（pumps）季對季增加了 25%。在原本就高度活躍的使用者基礎之上，這是非常大的互動變化。

And I think when you put the LinkedIn lesson up against Doximity, you realize that the larger and the more engaged your base of users is, the larger your monetization opportunities over a long period of time. That base is growing. It's growing rapidly. And we're just getting started on the monetization front, and I think we feel pretty excited about what's to come.

我認為，當你把 LinkedIn 的經驗放到 Doximity 來看，你會意識到：你的使用者基礎越大、參與度越高，長期的變現機會就越大。而這個基礎正在成長。它成長得很快。我們在變現面才剛起步，我想我們對接下來的發展感到相當興奮。

Michael Cherny, Leerink Partners.

Leerink Partners 的 Michael Cherny。

Congrats on the quarter. Maybe I want to take a step back on all things AI, but dive around the broader AI strategy beyond just AI search. And 1 area I've been thinking about is the work that you've done so far with Scribe. I know it's still an early launch. I know you're still early in the monetization efforts.

恭喜這一季的表現。也許我想先從 AI 的整體議題退一步看，但更深入談談不僅限於 AI 搜尋的更廣泛 AI 策略。我一直在思考的一個領域，是你們目前在 Scribe 上所做的工作。我知道這仍是早期推出階段。我也知道你們在變現方面仍在早期。

But can you give us a sense now on what you're seeing on subscribe either as an entry point as a connective tool or -- and if there's something else beyond scribe, I'm not asking about, but some other areas where pieces that you have in place can help bring the totality of the search function together to clients?

但你能否現在分享一下，你們在 Scribe（原文為 subscribe）上看到的情況：它是否可作為切入點、作為連結工具？或者——如果還有其他超出 Scribe 的部分，我不是在問我沒問到的東西——但是否有其他你們已具備的模組，能協助把搜尋功能的整體性整合起來提供給客戶？

Yes. Thanks, Michael. This is Jeff. I'll say we're very proud of our tenant scribe growth and actually been our fastest grower in terms of number of doctors per month picking it up. And a big part of that enterprise hospital motion. It sits right between our health, our dialer and our clinical decision support, ask in a really nice way. It is the connected glue. I believe that five years from now, every physician will have an effect a doctor's digital assistant.

是的。謝謝你，Michael。我是 Jeff。我想說，我們對我們的 tenant scribe 成長非常自豪，而且就每月採用的醫師數而言，它其實是我們成長最快的產品。其中很大一部分來自企業端醫院的推進動能。它很巧妙地位於我們的 health、我們的 dialer，以及我們的臨床決策支援 ask 之間。它是把一切串起來的黏著劑。我相信五年後，每位醫師都會有一個醫師的數位助理。

And the core of that doctor's digital assistant will be us, and it will be the combination of drive and ask our note-taking tool that then leads to a lit search or a clinical decision support tool. And that just makes sense. At the end of the day, when you want to ask questions after seeing a patient wondering what the treatment options might be or what might I have missed in the differential, it's easiest to do that, of course, when you have the whole visit and the pre charting all there.

而那個醫師數位助理的核心會是我們，會是 drive 與 ask 的組合：我們的筆記工具，接著導向文獻搜尋（lit search）或臨床決策支援工具。這樣的路徑非常合理。歸根結底，當你看完病人後想提問、想知道可能的治療選項是什麼，或在鑑別診斷上我可能漏了什麼時，最容易的方式當然是：整個看診內容與事前預先建檔（pre charting）都完整在那裡。

And of course, that's easiest to do when you're flowing out of the telehealth tool that you're using, that's already right there to take the notes and have that available for you. And it all leverages the enterprise relationships we built up over the years again, with all of the HIPAA security SOC 2 reviews, we're really a platform for our health system clients.

而這當然也最容易發生在你使用的遠距醫療工具流程之中：它本來就能在那裡做筆記，並把內容提供給你使用。而這一切也都運用到我們多年來建立的企業關係；再加上所有 HIPAA 安全性與 SOC 2 審查，我們對健康系統客戶而言，確實是一個平台。

We're not a point solution. And that's really important. Actually, health systems don't like buying point solutions. So tend to be security vulnerabilities for them and more overhead, honestly, than it's worth. But again, being a platform for them that can be the doctor's digital assistant, I think, is a really powerful place to be.

我們不是單點解決方案。這點非常重要。事實上，健康系統不喜歡購買單點解決方案。對他們而言，那往往是安全性弱點，而且老實說，額外的管理負擔可能大於其價值。但再次強調，作為能成為醫師數位助理的平台，我認為這是一個非常有力的位置。

If you look at the leaders in the scribe market today, it's Microsoft. If you look at the leaders in clinical decision support, it's up to date. Again, I think telehealth is the biggest connector between the two of those. And I think we're in a strong position here to, again, be that combined doctor's digital system. No one else is top three in both note-taking, scribe and top three in clinical decision support. We are.

如果你看目前 scribe 市場的領導者，是 Microsoft。如果你看臨床決策支援的領導者，是 UpToDate。同樣地，我認為遠距醫療是把這兩者連結起來的最大連接器。我認為我們在這裡處於強勢位置，能夠——再次——成為那個整合式的醫師數位系統。沒有其他公司同時在筆記／scribe 與臨床決策支援兩邊都名列前三。我們是。

Ryan Daniels, Blair.

Blair 的 Ryan Daniels。

Matt, I wanted to go back to something you said in the prepared comments, I halt was interesting. I think you indicated that the AI programs had shorter inventory counts, maybe just a quarter and as you're moving into the upfront, you're seeing longer contract terms.

Matt，我想回到你在事先準備的發言中提到的一點，我覺得很有意思。我想你提到 AI 專案的庫存量（inventory）上限較短，可能只有一季；而當你們進入 upfront（預售）時，你們看到更長的合約期限。

Is that you guys changing the terms to kind of push them longer given the innovations you're seeing? Or is that just the market more receptive to longer contract terms?

這是你們因為看到創新而調整條款、把期限推得更長嗎？還是只是市場對較長合約期限的接受度提高了？

Yes, I can start, and then I'll have Perry jump in with some extra detail. I think for us, it's a pretty conscious decision. We wanted to -- I think we talked about in the prepared remarks, both protect the user experience and also iterate on the product and the go-to-market on what's a new product, not just for us but for our customers. And so we had more conservative inventory caps.

是的，我可以先回答，然後再請 Perry 補充一些細節。我認為對我們來說，這是相當有意識的決定。我們希望——我想我們在準備好的發言中也談到——同時保護使用者體驗，並且在一個新產品上迭代產品與 go-to-market（上市／商業化）策略；這不只是對我們是新產品，對客戶也是。因此我們設定了較保守的庫存上限。

We had shorter-term contracts in place for the first preliminary couple of tranches of customers that we've onboarded. As we go into the upfront, we'll open that up more towards larger and larger and longer contracts. And there's some really great signs and signals that we're seeing around demand are, maybe you want to talk about like a few words and how that's evolving.

對於我們導入（onboard）的前幾批初期客戶，我們採用較短期的合約。當我們進入 upfront 時，我們會更開放，朝向更大規模、期限更長的合約。而我們在需求面也看到一些非常好的跡象與訊號；Perry，也許你可以談幾句它如何演進。

Yes, absolutely. Ryan, so just to clarify there, too. Yes, I mean, I think for next year -- this has proven to be a good test round ahead of the upfront, right? And I think as you go into upfront, the focus shifts towards like said longer contracts, larger contracts you kind of organically have as the product grows more inventory per therapeutic area to offer.

是的，當然。Ryan，我也再澄清一下。是的，我的意思是，對明年而言——這已被證明是在 upfront 之前很好的測試回合，對吧？而我認為當你進入 upfront，重點就會轉向如你所說：更長的合約、更大的合約；而且隨著產品成長，你會很自然地在每個治療領域提供更多庫存量。

And as you feel more comfortable with the user experience, you can kind of get a little less conservative with ad load or cats. So that's kind of a natural progression. Also, as we've been in market and we've taken in feedback, it's clear that there are many other therapeutic areas we hadn't even kind of come to market with that people want.

而當你對使用者體驗更有把握時，你就可以在廣告負載（ad load）或上限（caps）上不必那麼保守。所以這是一個自然的進程。另外，隨著我們在市場上運作並吸收回饋，也很清楚有許多我們甚至還沒真正推向市場的其他治療領域，是大家想要的。

And I think with my RevOps hat on, it's been really encouraging in my seat to see hundreds and hundreds of salesperson requests to price out new categories to kind of work in new ways to do deals, whether it be category-specific or specific key or to a specific target list.

而且我想以我戴著 RevOps（營收營運）帽子的角度來看，坐在這個位置上，看到數以百計、數以百計的業務人員提出需求，希望為新類別進行報價，以便用新的方式來做交易——不論是按類別、按特定關鍵字，或是針對特定目標名單——這都讓人非常振奮。

There's just a lot of engagement and activity and buzz, and quite frankly, kind of the velocity of the business that we haven't solved a few quarters. So I think it's a good time to kind of get the product down and get the experience down to then kind of sell the larger contracts ahead of upfront. And so then the focus will shift a little bit.

整體參與度、活動量和討論熱度都很高，坦白說，這種我們已經好幾個季度沒有解決的業務推進速度又回來了。所以我認為現在是個好時機，先把產品本身和使用體驗打磨到位，接著再去提前銷售更大額、期限更長的合約。因此接下來重點會稍微轉移一些。

And the only other thing I'd add is, I think it's just really smart to take a deliberate approach and build building a measured way to start and then ramp it up more as you learn. Again, new products for us, but also new products for our customers. And I think we're really starting to find a nice product market fit there.

我唯一還想補充的是，我認為採取審慎、有意識的方式很明智：先以可衡量、循序漸進的方式起步，然後在學習之後再加速擴大。畢竟這些對我們來說是新產品，對客戶來說也是新產品。而我覺得我們真的開始在那裡找到很不錯的產品市場契合度。

Got it. Super helpful. And then, Jeff, my follow-up is for you. You mentioned that just having the AI solutions is opening up new opportunities for you to present to leaders in the industry, and it sounds like it's really helping the broader base.

了解。非常有幫助。接著，Jeff，我的追問是給你的。你提到僅僅是擁有 AI 解決方案，就為你們打開了向產業領導者展示的新機會，而且聽起來它也確實在幫助更廣泛的客群基礎。

I'm curious if you're actually seeing an accretive effect outside of that with kind of your core legacy offerings where they want to kind of bundle AI search and the efforts there with more of the core programs or initiatives to kind of do a one plus one equals three type of HCP marketing initiatives? Any color there would be great.

我想知道，除了這點之外，你們是否也在核心的既有（legacy）產品上看到增益效應（accretive effect）——也就是他們希望把 AI 搜尋以及相關投入，和更多核心方案或計畫綁在一起，做出那種「一加一等於三」的 HCP（醫療專業人員）行銷推進？如果能分享一些細節就太好了。

Thanks, Ryan. Yes. Actually, you put you're finger right on it. Once you start talking about AI search and how it's great to see that this group of doctors has concerns about your side set profile or about how to convert your dosing from a competitive drug to your drug that actually creates an opportunity for us to then have additional use of our telehealth platform to remind them of the dosing changes or the side effect studies that have been done. So it actually adds more utility and greater signal for the rest of the platform.

謝謝你，Ryan。是的。其實你正好點到關鍵。一旦你開始談 AI 搜尋，並且看到例如某一群醫師對你們的副作用特徵（side effect profile）有疑慮，或是對於如何把用藥劑量從競品藥物轉換到你們藥物有問題，這就會為我們創造機會：進一步使用我們的遠距醫療平台，提醒他們劑量變更，或提醒已完成的副作用研究。所以它確實為平台其他部分帶來更高的效用與更強的訊號。

But you really need both. At the end of the day, having signal is useless unless you have reach and having reach isn't as effective if you don't have good signal. And so bringing the two together has been a real unlock for us across our core business and across the new AI business.

但你真的需要兩者兼具。歸根結底，有訊號（signal）但沒有觸及（reach）是沒有用的；而只有觸及但沒有好的訊號，效果也不會那麼好。因此把兩者結合，對我們在核心業務以及新的 AI 業務上，都是一個真正的突破。

Ryan MacDonald, Needham & Company.

Needham & Company 的 Ryan MacDonald。

Just maybe to start with you. You talked about 165 signed health system clients and that there's more of a sort of maturation and top-down approach within these health systems to sort of have something that's secure and sort of a top-down policy in terms of usage on clinical evidence.

我可能先從你開始問。你提到有 165 家已簽約的醫療體系客戶，而且在這些醫療體系內部，正出現更成熟、由上而下的做法——希望在臨床證據使用方面，有一套安全的、由上而下的使用政策。

Can you just talk about where we're at in terms of what percent of those 165 health systems are live? And what the sort of maturation process looks like in terms of kind of if you want to call it, policing rogue clinical evidence use amongst the physicians within those systems?

你能談談目前這 165 家醫療體系中，有多少比例已經正式上線（live）嗎？以及所謂的成熟化流程大概長什麼樣——如果可以這麼說的話——在這些體系內，如何去「管控」醫師之間未經授權、各自為政的臨床證據使用？

Yes, it's a great question. And it's actually right in the middle of what's the foot right now. So I'll begin by saying, we're live at all of those 165. And insofar as we have a privacy agreement in place and their doctors know that they're able to use our AI tools. We're in various stages of implementation of getting into their electronic health records and having the links and making all those things work.

是的，這是個很好的問題。而且它其實正好處在目前最核心的議題中。我先說結論：這 165 家我們全部都已經上線。就我們已經簽署隱私協議、且他們的醫師知道可以使用我們的 AI 工具這一點而言，都是上線狀態。至於把系統接入他們的電子病歷（EHR）、建立連結並讓所有功能順暢運作，我們正處於不同的導入階段。

And that's a big effort for our team, but it's again another area where we're going to invest heavily because we know from our telehealth experience that once you're into these systems, it's a very sticky place to be, and you're really part of the workflow. In terms of where they're at from just a legal perspective and looking at all this I would say the concern is slowly ramping.

這對我們團隊來說是很大的工程，但這同樣也是我們會大力投資的領域，因為我們從遠距醫療的經驗知道，一旦你進入這些系統，那會是一個非常「黏著」（sticky）的所在，你會真正成為工作流程的一部分。至於從法律角度以及他們如何看待這一切，我會說相關的擔憂正在逐步升溫。

I mean, you're seeing a few lawsuits here and there that start to play out. The most recent one was against open AI with an individual patient who had hordes of breadth, and it told him to wait it out, sitting his recliner chair for two days back and forth, and again, we'll see how this lawsuit plays out, but this patient had a pulmonary embolism, which should have brought him to the emergency room right away. And so those types of stories, I think, rightfully raise the hair and the neck of the AI steering committees and the MedMal committees.

我的意思是，你會看到零星出現一些訴訟開始發酵。最近的一起是針對 OpenAI：一位個別病患有呼吸急促（shortness of breath），系統卻告訴他撐一撐，坐在躺椅上來回等了兩天；當然我們還要看訴訟最後如何發展，但這位病患其實是肺栓塞（pulmonary embolism），本應該立刻去急診。所以這類故事，我認為理所當然會讓 AI 指導委員會（AI steering committees）以及醫療疏失（MedMal）委員會提高警覺。

And so when you're making high-stakes decisions like this, it's really important to have the right experts there, and it's hard to do. So I'll say this. When it comes to high stakes things, we have pilots that are landing planes in this country. We have doctors that treat patients I think we'll continue to see both those things continue to happen, pilot land plans, doctors treat patients.

因此，當你在做這種高風險決策時，讓合適的專家在場非常重要，而這也很難做到。所以我會這麼說：在高風險的事情上，我們國家有飛行員在降落飛機。我們也有醫師在治療病患，我想我們會繼續看到這兩件事持續發生——飛行員降落飛機、醫師治療病患。

And when the doctors treat patients at the hospitals, they're going to want to make sure that their patient data is safe and not being put out to some unauthorized AI. So on that front, I can tell you, the largest health system in the country, I know has blocked a bunch of AI players, not us. We're working with them. We're seeing I'd say, half a dozen others who've done similar.

而當醫師在醫院治療病患時，他們會希望確保病患資料是安全的，不會被送到某個未經授權的 AI。在這方面我可以告訴你，全美最大的醫療體系之一，我知道他們已經封鎖了不少 AI 業者——不是我們。我們正在與他們合作。我也看到大概還有半打其他醫療體系做了類似的事。

We haven't really done a survey of them per se, but I think you'll see, as the year progresses here, we're moving from this kind of age of AI adoption to AI accountability. And I think by this time next year, we'll definitely be looking at a place where the enterprise will be saying, no, this should be the AI you use.

我們並沒有特別對他們做一個正式調查，但我認為你會看到，隨著今年進展，我們正從這種「AI 採用」的時代，走向「AI 問責」的時代。而我想到了明年這個時候，我們肯定會進入一個階段：企業會說，不，你就應該使用這一個 AI。

And again, I liken this back to putting assistance in anthropic and OpenAI A year ago, it was really try out whatever helps you. Cursor and Claude Code and Codex and all of them. And then really somewhere around the beginning of this calendar year, you started to see enterprises saying, no, this is the one we prefer, and this is the one we're going to have a security agreement with and volume pricing with again, I think you'll see that similar shift to the enterprise happen as healthcare AI use becomes more than norm.

同樣地，我把這比作一年前企業在 Anthropic 和 OpenAI 等助理工具上的做法：當時基本上是「你就試試看任何能幫上忙的」。Cursor、Claude Code、Codex，全部都可以。但大概從今年曆年年初開始，你就看到企業開始說：不，我們偏好這一個；我們會跟這一個簽安全協議，也會談量價（volume pricing）。同樣地，我認為隨著醫療 AI 的使用變得不再只是新鮮事、而更接近常態，你也會看到醫療領域出現類似的企業端轉變。

Ryan, I'll just add one thing really quickly. Again, kind of plumbing in much more recently. I think this is where the NOHARM study is very validating for the approach that we've taken over the last year. in that it shows that Doximity is most accurate and safest clinical AI tool in the market today. And that's certainly what our kind of health systems customers are looking for.

Ryan，我再很快補充一點。同樣地，這些「底層管線」（plumbing）其實是更近期才逐步接上。我認為這正是 NOHARM 研究對我們過去一年所採取方法的強力驗證：它顯示 Doximity 是目前市場上最準確、也最安全的臨床 AI 工具。而這也正是我們醫療體系客戶所尋求的。

Our team should be really proud here of the work that they've done over the last year to kind of take what I think Doximity is calling card in a physician's first approach in developing the product. So I think that it's worth calling out because it's really relevant to the question you asked.

我們的團隊應該為過去一年所做的工作感到非常自豪——把我認為 Doximity 的招牌，也就是以醫師為先（physician-first）的方式，落實在產品開發上。所以我覺得值得特別點出來，因為它和你剛才問的問題高度相關。

That's really helpful color there. And as we think about sort of the next layer on to that and sort of monetizing AI search Will there be some sort of unlocking factor in terms of the revenue opportunity on AI search as sort of we get -- make progress with these health systems sort of rolling out Doximity ask more broadly and then seeing sort of that maybe more material ramp in usage or continuing with your ramp in usage? Is there sort of a knock-on benefit here as we think about sort of the AI search product and ability to sell it?

這些補充說明非常有幫助。那當我們再往下一層思考、也就是 AI 搜尋的變現時，隨著我們在這些醫療體系更廣泛地推動 Doximity Ask 的導入，並看到使用量可能更明顯地加速爬升、或延續目前的使用成長曲線，AI 搜尋在營收機會上是否會出現某種「解鎖」因素？從 AI 搜尋產品以及其銷售能力的角度來看，是否會有連帶的正向效益？

I mean, fundamentally, the AI search product is going to be based on the number of doctors who are using it. And so yes, that unlike will certainly happen.

我的意思是，從根本上來說，AI 搜尋產品會以使用它的醫師人數為基礎。所以是的，那確實會發生。

Elizabeth Anderson, Evercore ISI.

Elizabeth Anderson，Evercore ISI。

Appreciate all the color as you talk about the new opportunities that your -- the new AI launch has afforded you guys. I wanted to just think, if you think about your sort of pharma side of the customers, do you have any sense on how big those AI budgets are this year, for example, versus their traditional pharma ad budgets? Is it something that they're just maybe is 5% to 10% of the size?

感謝你們分享許多細節，談到這次新的 AI 發布為你們帶來的新機會。我想請教一下，從你們藥廠端客戶的角度，你們是否掌握今年他們的 AI 預算大概有多大，例如相較於他們傳統的藥品廣告預算？這會不會大概只有其規模的 5% 到 10%？

And then sort of have you had any preliminary conversations with customers about how they're growing that? Because obviously, that should increase over time, one would think, in terms of like TAM expansion of opportunity. But I just want to kind of here, like where we are today and where you guys think that we might be, say, next year?

另外，你們是否已經和客戶有一些初步對話，了解他們打算如何擴大這部分預算？因為很明顯，隨著時間推進，從可服務市場（TAM）擴張的角度來看，這應該會增加。但我想更具體地聽聽：我們今天大概在哪個位置？以及你們認為例如到明年可能會到哪裡？

It's a great question. Honestly, I wish I had a better crystal ball to give you an answer. I think if you asked the Board of Directors of the top 20 pharma companies, they wouldn't know exactly where their AR budgets are landing this year because it is such a dynamic space also hear that one of the analysts who covers us did do a survey of up pharma buyers and did ask them what percent of their overall budgets they expected to spend on AI and it was around 10% or less than 10% and then when asked who they would spend that AI budget with, we were the number one choice.

這是個很好的問題。老實說，我希望我有更好的水晶球可以給你答案。我想就算你去問前 20 大藥廠的董事會，他們今年的 AI 預算最終會落在哪裡也未必說得準，因為這是一個非常動態的領域。另外，我也聽說有一位覆蓋我們的分析師確實做過一份針對藥廠買方的調查，問他們預期會把整體預算的多少比例花在 AI 上，結果大約是 10% 或低於 10%；而當被問到會把這筆 AI 預算花在哪裡時，我們是第一名的選擇。

And I was really proud of that given that, that survey was done only 1.5 months or so after we first came out with an AI search product. So here, it is just a couple of months into being out in market, we are already the number one choice by a pretty wide margin in terms of where they would spend their AI this year. And I think that it boils down to us having built a trusted relationship with them over time, having proven our ROI and having a team that knows how to work within the regulatory framework that exists but also knows how to innovate.

我對此非常自豪，因為那份調查是在我們首次推出 AI 搜尋產品後大約 1.5 個月左右就完成的。所以在上市才短短幾個月，我們就已經以相當大的差距成為他們今年 AI 預算投放的第一選擇。我認為這歸因於我們長期建立的信任關係、已經證明的投資報酬率（ROI），以及一支既懂得在既有法規框架內運作、也懂得創新的團隊。

So I wish I had a better answer though. I can say the overall Google paid search spend by pharma is in the $14 billion range. I mean, it's a very big number, but that does include direct-to-consumer and some other categories that we probably won't play in.

不過我還是希望能有更好的答案。我可以分享的是，藥廠在 Google 付費搜尋上的總支出大約在 140 億美元的量級。這是一個非常大的數字，但其中也包含直接面向消費者（DTC）以及其他一些我們可能不會參與的類別。

But at the same time, I also think the ability to be more ingrained in the medical decision-making is a bigger opportunity that strips well beyond what people are using a Google search for today. So we think the TAM is very large. It's certainly as large as the market we're in to get into AI search.

但同時我也認為，更深入嵌入醫療決策流程的能力，是一個遠遠超出人們今天使用 Google 搜尋所能涵蓋的更大機會。因此我們認為 TAM 非常大。就進入 AI 搜尋而言，它至少和我們目前所處的市場一樣大。

Got it. And maybe as a follow-up, like many of your other products, you sort of launched at a particular gross margin. And I appreciate this one might be slightly different. But do you see anything that would necessarily like inhibit the gross margin line of this new offering from starting to trend up over the longer term?

了解。那再追問一下，像你們其他許多產品一樣，你們通常會在某個特定的毛利率水準推出。我理解這個產品可能略有不同。但你們是否看到任何因素，可能會抑制這項新產品的毛利率線在長期趨勢上往上走？

(inaudible) question. I realize is the future of AI, but thank you for trying.

(聽不清楚) 的問題。我知道這是 AI 的未來，但謝謝你嘗試。

No, all good. I think the short answer is no. I think in terms of our margins this year, it's really an intentional time sequencing, where we wanted to invest and support the kind of user engagement experience of clinical AI and scribe as we grow them.

不會，沒問題。我想簡短的答案是：不會。就我們今年的毛利率而言，這其實是刻意的時序安排；我們希望在擴大臨床 AI 與 scribe 的同時，先投入資源來支撐使用者互動體驗。

And then I think the revenue opportunity is there to offset that investment as we begin to ramp it up over time. Actually, I think we're pleased that we're starting to see some opportunities to cost optimize even as we make a much larger investment. Some of that's around how we service the product to our users. Some of it is around how we're running our operations, reserving capacity, better forecasting.

接著，隨著我們逐步拉升規模，我們相信營收機會足以抵消這些投資。事實上，我們也很高興看到，即便在加大投資的同時，仍開始出現一些成本優化的機會。其中一部分與我們如何為使用者提供產品服務有關。另一部分則與我們如何營運、預留產能、以及更好的需求預測有關。

Some of it is just how we're tuning and optimizing different AI models as well as building our own first-party AI tooling around the products. All of those, I think, will help us get more efficient over time. But at the end of the day, it's really about sequencing our revenue and ramping that up, which will, I think, help improve the margin profile.

還有一部分是我們如何調校並優化不同的 AI 模型，以及圍繞產品建立我們自有的第一方 AI 工具。我認為這些都會幫助我們隨時間推進變得更有效率。但歸根結底，關鍵在於營收的時序與拉升速度；我認為這將有助於改善毛利率結構。

Steve Valiquette, Mizuho Securities.

Steve Valiquette，Mizuho Securities。

I think I was also trying to hope here to dig in a little bit deeper on the reacceleration of your capture on the pharma digital marketing spend. So maybe more from just a numerical standpoint. I know it's hard to put numbers on it, but historically, you've obviously talked about the market for pharma digital marketing spend towards practitioners growing in that 5% to 7% range.

我也想更深入了解一下，你們在藥廠數位行銷支出上的市占率擷取重新加速的情況。也許從數字角度來看。我知道很難量化，但過去你們顯然提過，藥廠針對醫療從業人員（practitioners）的數位行銷支出市場，大概以 5% 到 7% 的速度成長。

And then aim a normal environment would maybe grow double that market rate Obviously, a lot of that was pushed the side up for the past year or so. But the things reaccelerating again, maybe just frame numerically where you think market growth is right now, it's hard to predict next year, so I want to ask that, or is it another way with your 7% revenue growth this quarter, do you think you're getting share -- gaining market share at that number or just growing in line with the reexpanding market?

而在正常環境下，你們可能會以大約市場成長率兩倍的速度成長。顯然，過去一年左右很多因素讓成長被壓抑。但現在又重新加速了。也許你可以用數字框架一下：你認為目前市場成長率大概是多少？我知道很難預測明年，所以我不問那個。或者換個問法：以你們本季 7% 的營收成長來看，你覺得這代表你們在拿市占（市占率提升），還是只是跟著重新擴張的市場同步成長？

Hopefully, that all makes sense.

希望我這樣問有說清楚。

Steven, yes, I think the -- relative to last quarter, I don't think our expectations for the HCP market growth have materially changed, stable in the mid-single digits. Obviously, we grew better than that in Q1. Our guidance is to grow in kind of in line with that for the rest of the year. A lot of our revenue for this year kind of dates back to last year, when we didn't have an AI monetization product in the upfront cycle and we will this year.

Steven，是的，我想——相較於上季，我們對 HCP 市場成長的預期並沒有出現實質變化，仍穩定在中個位數成長。顯然，我們在第一季的成長優於此水準。我們的指引是今年剩餘時間大致會以接近該水準成長。我們今年的許多營收其實可追溯到去年，當時我們在預售（upfront）週期中還沒有 AI 變現產品，而今年我們會有。

So obviously, we'll see how that goes. We're excited for the building pipeline there and the demand that we're seeing from customers. And I think that should portend well to us being able to outgrow the market again over time. I think the growth rate that we have this year feels probably like more of a blip of kind of timing more than anything else.

所以當然，我們會看看進展如何。我們對於正在建立的銷售管線以及客戶端看到的需求感到興奮。我認為這預示著，隨著時間推進，我們有能力再次跑贏市場成長。我覺得我們今年的成長率更像是一個時點因素造成的小波動，而不是其他原因。

Jessica Tassan, Piper Sandler.

Jessica Tassan，Piper Sandler。

Congratulations on a really strong print and Matt on a really strong first quarter as CFO. So I guess, maybe, Jeff, I think you mentioned the strength in both hospital and pharma customers on just maybe any color on -- or directional commentary on kind of hospitals versus pharma? Are deployments able to monetize AI search in both of those avenues? Or is it kind of one or the other?

恭喜你們交出非常強勁的成績，Matt 也恭喜你作為 CFO 的第一季表現非常出色。我想問的是，Jeff，我記得你提到醫院與藥廠客戶兩邊都很強勁；能否再補充一些——或方向性的評論——關於醫院相較於藥廠的情況？在這兩條路徑上，部署是否都能讓 AI 搜尋變現？還是說主要會偏向其中一邊？

If you could just help us understand the hospital versus pharma split on AI search would be helpful.

如果您能協助我們理解在 AI 搜尋方面，醫院端與製藥端的占比拆分，會很有幫助。

Yes. We're doing well with both. So I'd say both are doing equally well in the hospital and pharma side of the business. I will say the budget source are slightly different on each. In the pharma side, this is generally creating a new budget. And so there is a bit of searching around for this new budget that didn't exist a year or two ago. But again, it's good ROI. It's good upside, and it's the sort of innovation that I know has the attention of the C-suite.

是的。我們在兩邊的表現都很好。所以我會說，在醫院端與製藥端的業務上，兩者表現同樣出色。我會補充一點：兩邊的預算來源略有不同。在製藥端，這通常是在建立一個新的預算。因此，會需要花一些時間去尋找這筆一兩年前還不存在的新預算。但再說一次，它的投資報酬率（ROI）很好。上行空間也不錯，而且這類創新我知道已經引起高層（C-suite）的關注。

Within hospitals, it's actually sometimes a cost savings because they have been spending a lot of money on clinical decision support and on scribe services from other vendors and we're coming in with an offering that's lower cost than what they may have been using historically. And so there's an opportunity for them to save some money.

在醫院端，有時其實是成本節省，因為他們過去在臨床決策支援以及來自其他供應商的醫療書記（scribe）服務上花了很多錢，而我們提供的方案成本低於他們歷史上可能使用的方案。因此，他們有機會省下一些費用。

And so that it becomes more of a budget opportunity for them to choose to work with us. And we're pleased to help them make that change. And again, our margins, as Matt described earlier, I think, can support that quite well. So on both fronts, on both hospital and pharma, we're seeing, again, high interest and growth on the AI products.

因此，這對他們而言更像是一個預算上的機會，讓他們選擇與我們合作。我們也很樂意協助他們完成這項轉換。而且如 Matt 先前所描述，我認為我們的毛利率也能很好地支撐這件事。所以在兩個面向上，不論是醫院端或製藥端，我們都再次看到對 AI 產品的高度興趣與成長。

Jessica, one other thing I'd add is just kind of next back to an earlier part of this conversation, which is talk about unlocking AI budget with pharma customers in particular. And I think that's true. I would also say that across a lot of the conversations that we're having, and Jeff talked about, sitting in new boardrooms to the ones we've had in the past.

Jessica，我還想補充一點，呼應我們先前談話較前面的部分，也就是特別針對製藥客戶如何解鎖 AI 預算。我認為這確實如此。我也想說，在我們正在進行的許多對話中——Jeff 也提到——我們走進了一些新的董事會會議室，和過去接觸的不太一樣。

There's a diversity of funding sources, I think, for marketing and AI search, right? Part of its AI innovation and that's new. And sometimes that feels sexy. I think there's also some of an age old insights and analytics budget that we haven't had opportunity to serve before that we're now able to, with some of the relevant that we can generate in the ad product.

我認為在行銷與 AI 搜尋方面，資金來源是多元的，對吧？其中一部分來自 AI 創新，這是新的。有時候這看起來很吸睛。我也認為，還有一部分來自傳統的洞察與分析（insights and analytics）預算，過去我們沒有機會服務，但現在我們能透過廣告產品所產生的一些相關性來切入。

And then also search, which Jeff talked about, right? Again, not as much a budget that's been available to us that I think will be more so going forward. And that's exciting. It's exciting to have, I think, a diversity of ways that we can grow going forward rather than just be dependent on kind of one kind of one new spending area that is moving very quickly.

另外還有搜尋（search），正如 Jeff 所談到的，對吧？同樣地，這並不是過去我們能取得的預算來源，但我認為未來會更有機會。這令人振奮。我覺得令人興奮的是，我們未來能以多元方式成長，而不是只依賴某一種、變化非常快速的新支出領域。

That's really helpful. And then my follow-up, you all have highlighted the drug reference component of your AI model. And I know, Jeff, you've got a long history with these types of products. So it seems like the drug reference is a really important differentiator within Doximity Ask.

這真的很有幫助。接著我的追問是，你們一直強調 AI 模型中的藥品參考（drug reference）組件。而且我知道 Jeff，你在這類產品上有很長的經驗。所以看起來藥品參考在 Doximity Ask 裡是一個非常重要的差異化因素。

So can you just give us some color on what is proprietary or important here, what should investors know about the drug reference tool? And is it uniquely well positioned to kind of to host sponsored content or just any detail or color that you'd like investors to understand about the drug reference component of the model?

所以你能否多分享一些細節：這裡有哪些是專有的或特別重要的？投資人應該了解藥品參考工具的哪些重點？以及它是否在承載贊助內容方面具有獨特優勢？或任何你希望投資人理解的、關於模型中藥品參考組件的補充說明？

Thanks, Jessica. Yes, certainly, I spent 11 years working on a drug reference company. So I can say that drug reference is a big part of clinical decision support, it's no mistake at all that Up To Date really didn't take off in the marketplace until they acquired Lexicomp, which was a drug reference and put those two together. And I think in a similar way, our Doximity Ask product, again, having a built-in drug reference is a key thing. And having it available is discrete data elements.

謝謝你，Jessica。是的，當然，我曾在一家藥品參考公司工作了 11 年。所以我可以說，藥品參考是臨床決策支援的重要部分；UpToDate 在市場上真正起飛，直到他們收購了藥品參考產品 Lexicomp 並把兩者結合在一起，這絕非偶然。我認為以類似的方式，我們的 Doximity Ask 產品同樣如此：內建藥品參考是一個關鍵。而且它是以離散資料元素（discrete data elements）的形式提供。

So basically, the AI can work affluently with the drug reference to have the right dose, appear with the right indication to do the drug-drug interaction check, so being able to do all that is one of the things that leads to more accurate results. And again, one of the reasons why we believe we won the NOHARM study.

基本上，AI 可以與藥品參考順暢協作，讓正確劑量能在正確適應症下呈現，並進行藥物—藥物交互作用檢查；能做到這些，是提升結果準確度的因素之一。而且這也是我們相信自己能在 NOHARM 研究中勝出的原因之一。

The other thing I'll highlight again, which plays into all of this is we're the only company in the industry that we're aware of that has really leaned into physician oversight to have these 12,000 physicians cited authors who are reviewing the AI outputs on a consistent basis and even layering on their expert views on top of what might be out in the published pubmed evidence world, you really do need to have some of that in some situations because let's face it, there isn't an RCT, a randomly controlled trial that is done in every type of morbidity in question.

我還要再強調另一點，這也與整體息息相關：就我們所知，我們是業界唯一真正投入醫師監督（physician oversight）的公司，擁有 12,000 名醫師擔任具名作者（cited authors），持續性地審閱 AI 輸出，甚至在已發表的 PubMed 證據基礎上再疊加他們的專家觀點；在某些情境下，你確實需要這樣做，因為坦白說，不可能針對每一種所討論的疾病狀況都做過 RCT（隨機對照試驗）。

And so we're able to see what questions are being asked most frequently and being able to provide answers to questions even if there isn't a clear question or clearing answer available on the Internet or in published journals. So anyway, we're really proud of how this has come together in the fusion of Pure Check, drug reference and AI evidence.

因此，我們能看到哪些問題被最頻繁地提出，並且即使在網路上或已發表期刊中沒有明確問題或明確答案可循時，也能提供問題的解答。總之，我們對於 Pure Check、藥品參考與 AI 證據的融合能以這樣的方式整合在一起，感到非常自豪。

And we think this is the reason why we're going to continue to win these independent studies looking at what's the highest clinical accuracy.

我們也認為，這就是我們將能持續在這些獨立研究中勝出的原因——這些研究會評估誰具有最高的臨床準確度。

Sean Dodge, BMO Capital Markets.

BMO Capital Markets 的 Sean Dodge。

Maybe just going back to the HCP marketing. You talked about the dynamics in large pharma and change in buying behaviors there. It sounds like that's getting better in places. But just curious on the SMB side. I know you've been able to drive a lot of share gains or growth in that part of the market with the portal. What percent of revenue now comes from like SMB pharmas now?

也許回到 HCP 行銷這一塊。你們談到大型藥廠的動態以及其採購行為的改變。聽起來某些地方正在改善。但我想問的是 SMB 端。我知道你們透過入口網站（portal）在那部分市場取得了不少市占提升或成長。目前營收中大約有多少比例來自 SMB 藥廠？

And have you seen kind of similar dynamics in buying play out in that part of the market, too?

而且你們是否也看到在那部分市場中，類似的採購動態也在發生？

Yes, it's Perry. I'm happy to take the question. So SMB has been really strong for us. I think traditionally, if you look at the investor deck our penetration with the top 20 is much higher than with SMBs. I think with all of the brands that are under $100 million. I think the last step we put out, we're around 10% of them we work with. There's a huge opportunity there it's been turbocharged by some of these kind of independent agency partnerships we add in the portal.

是的，我是 Perry。我很樂意回答這個問題。SMB 對我們來說一直非常強勁。我想傳統上，如果你看投資人簡報（investor deck），我們在前 20 大客戶的滲透率遠高於 SMB。以所有品牌年銷售額低於 1 億美元的來看，我記得我們上次公布的數據是，我們大約與其中 10% 合作。那裡有很大的機會，而且透過我們在 portal 中新增的一些獨立代理商合作夥伴關係，這個機會被大幅加速。

I can to that team that did really well this past quarter, they grew over 100%. So they're doing really well. I think AI search is also kind of providing more opportunity for that team to get in with certain companies that may not has already gravitated towards other products in the past that are really interested in this. So it's kind of expanding their aperture.

我可以告訴你，那個團隊在上一季表現非常好，成長超過 100%。所以他們做得很出色。我認為 AI 搜尋也為那個團隊帶來更多機會，能切入某些公司——這些公司過去可能已經更傾向其他產品，但現在對這項產品非常感興趣。因此，這等於擴大了他們的視野範圍。

And so I think that part of the business has been really up. It's I think something we're super excited about. And it's diversifying our base of business in a way that we haven't seen historically.

所以我認為那部分業務確實大幅上升。我想這是我們非常興奮的一件事。而且它正以我們過去未曾見過的方式，讓我們的業務基礎更加多元化。

Alexei Gogolev, JPMorgan.

Alexei Gogolev，摩根大通。

Jeff, I was wondering if you could maybe comment on how you're packaging AI search today? Is it category bundles versus keyword-specific or maybe target let based and what do you expect to change in the upfront versus initial couple of (inaudible) cohort?

Jeff，我想請你評論一下你們目前是如何打包（銷售）AI 搜尋的？是以類別套裝為主、還是以特定關鍵字為主，或是以目標名單為基礎？另外，你預期在預售期（upfront）相較於最初幾個（聽不清）批次（cohort）會有哪些變化？

Thanks, Alexei, this is Jeff. Actually, Perry is the expert here. I'll let him answer.

謝謝你，Alexei，我是 Jeff。其實 Perry 才是這方面的專家。我讓他來回答。

Alexei, great question. So I think what I can say is when we started this was a therapeutic category base and there's a share of voice model. I think as time has gone on, we're trying to make it as easy to buy as possible. We want to broaden the customer base. And a lot of folks kind of jumped right on for share of voice in seat category.

Alexei，問得很好。我想我可以說的是，我們一開始是以治療領域（therapeutic category）為基礎，並採用聲量份額（share of voice）模式。隨著時間推進，我們正努力讓購買流程盡可能簡單。我們希望擴大客戶基礎。而且很多客戶一開始就直接在某些治療類別上採用聲量份額。

But as we learn and we have more folks kind of inquire, I think there's opportunities to kind of layer in some target list buys opportunity to let people adjust some of the therapeutic areas I think kind of add or take out keywords. Some of them are forced to, we call it suppressive hers from a label review. So I think we are allowing for more and more options to kind of interact with the platform, and that is opening up more deal opportunity and driving more business.

但隨著我們學習、也有更多客戶來詢問，我認為有機會再疊加一些目標名單（target list）的購買選項，讓客戶能調整某些治療領域，並可新增或移除關鍵字。其中有些是被迫的，我們稱之為因標籤審查而進行的抑制（suppressive）。因此我認為我們正在提供越來越多與平台互動的選項，這也帶來更多成交機會並推動更多業務。

So I think we very much want to find the right product market fit and we are expanding the ways in which people can buy from us in one of the other unique things about the product as it exists today that will continue to iterate and experiment with is that there's a, I guess, we call it retargeting, but really an ancillary buy that you could do back into the core platform in parallel with the purchase that you do in AI search.

所以我認為我們非常希望找到合適的產品市場契合（product-market fit），並擴大客戶向我們購買的方式。

So that, I think, can be really powerful as an incremental and large purchase that really supports the core business above and beyond the new product. And to double click on that, so how do you set and monitor AI search inventory caps and ad load to protect the clinical experience and what quantitative signals tell you it's safe to open more inventory.

所以我認為，這作為一筆增量且可能規模很大的採購非常有力，能在新產品之外進一步支撐核心業務。再深入追問一下：你們如何設定並監控 AI 搜尋的庫存上限（inventory caps）與廣告負載（ad load），以保護臨床使用體驗？以及有哪些量化訊號能告訴你們可以安全地開放更多庫存？

Carefully. I mean, I think this is honestly one of the things I love about coming to Doximity is the amount of care that's put into kind of protecting and creating a really high-quality user experience. And so that doesn't even -- I think we're really conscious about inventory caps, yes, but we're also conscious about the quality of the ad. And is it helpful to doctors as it comes into the user experience.

謹慎地。我的意思是，我認為這老實說是我喜歡加入 Doximity 的原因之一：我們非常重視保護並打造高品質的使用者體驗。所以這不只是——我想我們確實很注意庫存上限，是的，但我們也同樣重視廣告的品質。以及它在使用者體驗中呈現時，是否對醫師有幫助。

So yes, cheerfully to Jeff's point. But there's a lot of thoughtfulness and a lot of, I think, rules and analysis that goes into getting that balance is right.

所以是的，呼應 Jeff 的觀點。但在取得這個平衡上，我們投入了很多周全思考，也有很多我認為的規則與分析來確保拿捏得當。

Alexei, I'll give you a quick parallel in the early days of telehealth. We had very, very conservative inventory caps even at a doctor that was on the product all day, all week, we've still shown on at every 4 or 5 days. As time goes on, utilized they love the product. It feels the ad feels native to the experience at value add for them, there's a little bit more room to loosen those caps.

Alexei，我給你一個早期遠距醫療（telehealth）的類比。當時我們的庫存上限非常、非常保守；即便是一位整天、整週都在使用產品的醫師，我們也可能每 4 到 5 天才展示一次。隨著時間推進，使用率提升，他們也喜歡這個產品。廣告的感受很原生、融入體驗，並且對他們有價值，那麼就會有更多空間去放寬這些上限。

But you cannot fit very carefully. But as time goes on, I think there'll be even more inventory, not just from atlas a little bit less restrictive, but also just more inventory with therapeutic areas.

但你必須非常謹慎地調整。不過隨著時間推進，我認為庫存會更多，不只是因為限制稍微放寬，也因為治療領域本身會帶來更多庫存。

Scott Schoenhaus, KeyBanc.

Scott Schoenhaus，KeyBanc。

Just wondering how AI monetization will impact sort of traditionally, you're not -- I don't want to call it lumpiness, but you used to see a lot of upfront and then renewals when you see midyear budget unlock. I mean, does this sort of smooth out the budget throughout the year? Can you just walk me through how this changes sort of any seasonality from your legacy platform?

我想了解 AI 變現會如何影響——傳統上，你們的營收並不是——我不想稱之為波動性（lumpiness），但過去常看到預售期（upfront）有一大部分，然後在年中預算解鎖時出現續約。我的意思是，這是否會讓全年預算分布更平滑？你能否帶我了解一下，這會如何改變你們既有平台的季節性？

Scott, this is Matt. Good actually a really good question. The -- I think a couple of thoughts. I mean, as it sits now, the buying mechanics aren't drastically different than the business has historically been, which is you buy on contract I think this will be sold. I think AI search will be sold considerably during the upfront season, the contracts last different periods over time.

Scott，我是 Matt。這確實是個非常好的問題。我有幾點想法。就目前而言，購買機制與我們歷來的業務並沒有大幅不同，也就是以合約方式購買，我認為這會以銷售合約的形式進行。我認為 AI 搜尋在預售期（upfront season）會賣出相當多，而合約期限會隨時間而有所不同。

If I -- we've talked a lot about my LinkedIn experience. And I think one of the most powerful pieces of it was that LinkedIn evolved and went into a more kind of biddable option-based world over a long period of time. It didn't happen overnight. And by doing that, and unlock the mass of business that kind of grew from a few hundred million dollars and $1 billion to $10 billion.

如果我——我們談過很多我在 LinkedIn 的經驗。我認為其中最有力的一點是：LinkedIn 在很長一段時間內逐步演進，走向更偏競價（biddable）、以選項為基礎的模式。這不是一夕之間發生的。而透過這樣做，解鎖了大量業務，讓規模從幾億美元、10 億美元成長到 100 億美元。

I think that potential and opportunity is in front of us, it's not something that will evolve and change overnight, a different way of buying is not something that we're mailing this year necessarily. But it is out there as an opportunity set for us. And I think that's a good thing. It's a good thing for our customers. And I think it would be potentially a very good thing for our business as well to kind of continue to evolve that buying model.

我認為這樣的潛力與機會就在我們眼前，但它不會一夜之間演進與改變；不同的購買方式也不一定是我們今年就要推行的。但它確實是我們的一個機會集合。我認為這是件好事。這對我們的客戶是好事。而且我認為，對我們的業務而言，持續演進這個購買模型也可能會非常有利。

Ladies and gentlemen, that concludes our question-and-answer session. I will now turn the conference back over to Mr. Jeff Tangney for closing remarks.

各位女士、先生，問答環節到此結束。接下來我將把會議交回 Jeff Tangney 先生作結語。

Thanks. I'd just like to end by thanking the entire Doximity team for the hard work serving more doctors every day than ever before. So thank you, everyone, for joining. Bye now.

謝謝。最後我想感謝整個 Doximity 團隊的辛勤付出，每天服務的醫師數量都比以往更多。也謝謝各位的參與。再見。

And ladies and gentlemen, this concludes today's call, and we thank you for your participation. You may now disconnect.

各位女士、先生，今天的電話會議到此結束，感謝各位的參與。您現在可以掛線。