SoundHound AI Inc (SOUN) 2025 Q3 法說會逐字稿

Good day, and thank you for standing by. Welcome to SoundHound's third-quarter 2025 earnings conference call. (Operator Instructions) Please be advised that today's conference is being recorded.

您好，感謝您的耐心等待。歡迎參加 SoundHound 2025 年第三季財報電話會議。（操作人員指示）請注意，今天的會議正在錄音。

I would now like to hand the conference over to your first speaker today, Scott Smith, Head of Investor Relations. Please go ahead.

現在我謹將會議交給今天的第一位發言人，投資者關係主管史考特史密斯。請繼續。

Good afternoon, and thank you for joining our third-quarter 2025 conference call. With me today is our CEO, Keyvan Mohajer; and our CFO, Nitesh Sharan. We will begin with some short remarks before moving to Q&A.

下午好，感謝各位參加我們2025年第三季電話會議。今天陪同我的是我們的執行長 Keyvan Mohajer 和我們的財務長 Nitesh Sharan。在進入問答環節之前，我們將先作一些簡短的發言。

We would also like to remind everyone that we will be making forward-looking statements on this call. Actual results could differ materially from those suggested by our forward-looking statements. Please refer to our filings with the SEC for a detailed discussion of the risks and uncertainties that could affect our business, and for discussion statements that qualify as forward-looking statements.

我們也要提醒大家，本次電話會議中我們將發表一些前瞻性聲明。實際結果可能與我們前瞻性聲明中暗示的結果有重大差異。有關可能影響我們業務的風險和不確定性的詳細討論，以及構成前瞻性陳述的討論性聲明，請參閱我們向美國證券交易委員會提交的文件。

In addition, we may discuss certain non-GAAP measures. Please refer to today's press release for more detailed financial results and further details on the definitions, limitations, and uses of those measures and reconciliations from GAAP to non-GAAP. Also note that the forward-looking statements on this call are based on information available to us as of today's date. We undertake no obligation to update any forward-looking statements except as required by law. Finally, this call is being audio webcast in its entirety on our Investor Relations website, and audio replay will be available following today's call.

此外，我們可能也會討論一些非GAAP指標。有關更詳細的財務業績以及這些指標的定義、限制和用途以及從 GAAP 到非 GAAP 的調節的更多詳細信息，請參閱今天的新聞稿。另請注意，本次電話會議中的前瞻性陳述均基於截至今日我們所掌握的資訊。除法律要求外，我們不承擔更新任何前瞻性聲明的義務。最後，本次電話會議將在我們的投資者關係網站上進行全程音訊網路直播，會議結束後將提供音訊回放。

With that, I would like to turn the call over to our CEO, Keyvan Mohajer. Please go ahead, Keyvan.

接下來，我將把電話交給我們的執行長凱文·莫哈傑爾。請繼續，凱文。

Thank you, Scott, and thank you to everyone for joining the call today. Q3 marks another quarter of precise execution against our plan. Enterprise AI adoption is booming globally, and SoundHound is strengthening its leading position by anchoring its deployment in millions of endpoints across highly diversified industries and customers, with much more potential remaining in the near and long-term horizon.

謝謝斯科特，也感謝今天所有參加電話會議的朋友。第三季標誌著我們又一個季度精準地執行了計劃。全球企業人工智慧應用蓬勃發展，SoundHound 透過將其技術部署到各行各業數百萬個終端，鞏固了其領先地位，並且在近期和長期內仍有巨大的發展潛力。

In just the first three quarters of the year, we have already achieved a record year in revenue of $114 million, up 127%, and we are raising our outlook once again. This quarter, we also celebrated our 20th year as a company. We started in a Stanford dorm room with the mission to voice enable the world with conversational intelligence. Our long-term focus, dedication, and strategic execution has paid off, as the opportunities before us are now advancing at an exponential rate.

今年前三個季度，我們的營收已創下歷史新高，達到 1.14 億美元，成長 127%，因此我們再次上調了業績預期。本季度，我們也慶祝了公司成立20週年。我們最初在史丹佛大學的一間宿舍裡起步，目標是利用對話智能賦能世界。我們長期的專注、奉獻和策略執行已經取得了回報，因為我們面前的機會現在正以指數級的速度發展。

Two decades of technical innovation has given us the speed and agility to truly capitalize on these opportunities. Our deep understanding of AI has allowed us to achieve market readiness rapidly, while many others are still experimenting. Indeed, we are already in the market achieving real success and creating measurable value for our customers with our technology as the driving force. And we have a track record of groundbreaking work and being highly responsive to new technological advancements.

二十年的技術創新使我們擁有了真正掌握這些機會的速度和靈活性。我們對人工智慧的深刻理解使我們能夠迅速實現市場準備，而許多其他公司仍在進行試驗。事實上，我們已經進入市場，取得了真正的成功，並以我們的技術為驅動力，為客戶創造了可衡量的價值。我們擁有開創性的工作記錄，並且能夠對新技術進步做出快速反應。

We pioneered speech-to-meaning over 10 years ago, which combines speech recognition and language understanding in real time to deliver superior speed and accuracy. Likewise, we pioneered deep meaning understanding over 10 years ago, and we're the first voice technology company to enable the processing of complex and compound conversations while others were still delivering simple limited commands. Our work a decade ago paved the way to leading the world in the agentic experiences we are seeing today.

十多年前，我們率先推出了語音轉語意技術，該技術將語音辨識和語言理解即時結合起來，從而實現了卓越的速度和準確性。同樣，我們在 10 多年前就率先實現了深度含義理解，我們是第一家能夠處理複雜複合對話的語音技術公司，而其他公司當時還在提供簡單的有限命令。十年前我們的工作為我們今天在智慧體驗領域引領世界鋪平了道路。

We were the first to go into production with a voice-enabled generative AI assistant in automotive within weeks of LLMs becoming a proven architecture in language understanding and years ahead of big tech. and this year, we were one of the first in the world to introduce a fully agentic platform for enterprise businesses, Amelia 7. Importantly, thanks to our years of IP accumulation and our mature platform, we are able to combine deterministic flows with machine learning models where the latter still lacks the reliability to go from proof of concept into production. This advantage has enabled us to deploy faster and scale faster and also avoid the explosion in model cost that other companies are battling today.

在語言學習模型（LLM）成為成熟的語言理解架構後的幾週內，我們就率先將支援語音的生成式人工智慧助理投入汽車行業的生產，並領先大型科技公司數年。今年，我們又成為全球首批推出企業級業務的全智能體平台 Amelia 7 的公司之一。重要的是，由於我們多年的智慧財產權累積和成熟的平台，我們能夠將確定性流程與機器學習模型相結合，而後者目前仍缺乏從概念驗證到生產的可靠性。這一優勢使我們能夠更快地部署和擴展，並且避免了其他公司目前正在努力應對的模型成本爆炸性增長。

Now let me give you an update on Polaris, our most recent groundbreaking work. SoundHound's multimodal, multilingual foundation model, Polaris, continues to prove its superiority in accuracy, speed, and cost. As we move more Amelia platform clients from third-party legacy vendors to use Polaris, we are able to reduce the error rate by as much as three times. Our customers are thrilled. and transitioning them to Polaris will help us drive down engineering and hardware costs and achieve faster iteration cycles to improve our speech foundation models.

現在讓我向大家介紹一下我們最新的突破性成果—北極星計畫。SoundHound 的多模態、多語言基礎模型 Polaris 不斷證明其在準確性、速度和成本方面的優越性。隨著我們將越來越多的 Amelia 平台客戶從第三方傳統供應商遷移到使用 Polaris，我們能夠將錯誤率降低多達三倍。我們的客戶非常興奮。將他們過渡到 Polaris 將幫助我們降低工程和硬體成本，並實現更快的迭代周期，從而改進我們的語音基礎模型。

We also continue to ask support for new languages and have innovated new methods to provide speech model customization with a rapid turnaround time and minimal deployment overhead. We've innovated new training methodologies that reduce the cost of training and amount of data required while significantly improving model accuracy.

我們也持續尋求對新語言的支持，並創新了新的方法，以快速的周轉時間和最小的部署開銷提供語音模型客製化服務。我們創新了新的訓練方法，在顯著提高模型準確性的同時，降低了訓練成本和所需資料量。

Notably, we are one of the very few companies that can support our customers in the environment of their choice, whether it is in our cloud, in their cloud, on the edge, or on-prem, as well as a hybrid combination. Polaris, in our view, is another significant disruption that widens the gap between us and the competition in our journey to realize SoundHound's vision. And while innovation is clearly a major source of strength and the foundation of our growth, We've also demonstrated repeatable success with our acquisition strategy.

值得一提的是，我們是極少數能夠支援客戶在其選擇的環境中使用的公司之一，無論是在我們的雲端、在他們自己的雲端、在邊緣上、在本地，還是在混合組合中。我們認為，Polaris 是另一個重大的顛覆性因素，它拉大了我們在實現 SoundHound 願景的道路上與競爭對手之間的差距。雖然創新顯然是我們力量的主要來源和成長的基礎，但我們的收購策略也取得了可重複的成功。

Within 12 to 18 months of our key acquisitions to date, we were able to convert their pre-measure decline to post-measure growth and turn them into leaders in their field as a fully integrated business unit within SoundHound. While acquisition is not a requirement for our success, it provides a unique opportunity for SoundHound to change the equation and accelerate our trajectory.

在迄今為止的關鍵收購完成後的 12 到 18 個月內，我們成功地將它們在收購前的下滑趨勢轉變為收購後的成長趨勢，並將它們轉變為 SoundHound 內部一個完全整合的業務部門，使其成為該領域的領導者。雖然收購並非我們成功的必要條件，但它為 SoundHound 提供了一個獨特的機會，可以改變現狀並加速我們的發展軌跡。

They've been able to find great businesses with amazing teams, strong customer relationships, and solutions highly aligned with our three pillars of business and arm them with what they needed to thrive, including SoundHound's strong IP, replacing their legacy tech dependencies with SoundHound in-house models that are more accurate, faster, and less costly, with 20 years of data and innovation behind them, improving their customer experiences while reducing their costs, strong financial backing for innovation and expansion, and proven scale, strong brand, and credibility.

他們找到了擁有優秀團隊、強大客戶關係和與我們三大業務支柱高度契合的解決方案的優秀企業，並為他們提供了蓬勃發展所需的資源，包括 SoundHound 強大的知識產權，用 SoundHound 更準確、更快速、成本更低的內部模型取代他們傳統的技術依賴，這些模型背後有 20年的數據和創新支持，在降低成本的同時改善客戶體驗，為創新和擴張提供強大的資金支持，以及經過驗證的規模、強大的品牌和信譽。

This quarter, we acquired Interactions, a pioneer in customer service and workflow orchestration, and we are already moving fast to combine functions to create a comprehensive and dynamic contact center and customer service offering that incorporates a full spectrum of automation and human-assisted capabilities. We have now demonstrated our M&A playbook multiple times, learning from each acquisition and getting faster and stronger every time.

本季度，我們收購了客戶服務和工作流程編排領域的先驅 Interactions，我們正在快速推進各項功能的整合，以創建一個全面、動態的聯絡中心和客戶服務產品，該產品融合了全方位的自動化和人工輔助功能。我們已經多次展示了我們的併購策略，從每一次收購中吸取經驗教訓，並且每次都變得更快更強。

Just as we did with our past acquisitions, we are now integrating our strong IP and replacing their dependencies on third-party models with more accurate, faster, and less costly SoundHound-built models. And with our robust financial position, we can give them the resources they need for innovation and expansion. With this latest combination, we expect to achieve the results we achieved through previous acquisitions. harvesting cost synergies by moving their stack into our own cloud and realizing revenue synergies with cross-selling and upselling.

正如我們過去收購其他公司時所做的那樣，我們現在正在整合我們強大的智慧財產權，並用更準確、更快速、成本更低的 SoundHound 自研模型取代它們對第三方模型的依賴。憑藉我們雄厚的財務實力，我們可以為他們提供創新和擴張所需的資源。透過此次最新合併，我們期望能夠取得與先前收購相同的成果：透過將他們的技術堆疊遷移到我們自己的雲端來獲得成本協同效應，並透過交叉銷售和向上銷售實現收入協同效應。

With that, let me now talk about some specific customer highlights in Q3. In IoT and Robotics, we had a significant win, signing a deal with a large Chinese company that offers intelligence-based interaction in hardware and software products. We agreed to integrate SoundHound Chat AI into double-digit millions of AI-enabled smart devices which will initially be distributed in the Indian market, leveraging our strong language capabilities in Indian languages.

接下來，我將重點介紹第三季的一些客戶亮點。在物聯網和機器人領域，我們取得了重大勝利，與一家大型中國公司簽署了協議，該公司在硬體和軟體產品中提供基於智慧的互動。我們同意將 SoundHound Chat AI 整合到數千萬台支援人工智慧的智慧型裝置中，這些裝置最初將在印度市場分銷，利用我們在印度語言方面的強大語言能力。

In Automotive, we continue to see strong adoption and have begun to deepen our market penetration beyond global light vehicles. We are excited to now be working with a major globally renowned sports car brand to develop a unique personality for its vehicle assistance. Additionally, Jeep vehicles rolled out our category-leading chat AI voice assistant in Europe, and our work with existing EV customers, including Lucid, Togg, and others, are seeing promising results. Togg has just recently expanded throughout the German market.

在汽車領域，我們持續看到強勁的市場接受度，並已開始加深我們在全球輕型汽車以外的市場滲透。我們很高興能與全球知名的跑車品牌合作，為其車輛輔助系統打造獨特的個性。此外，Jeep 汽車在歐洲推出了我們領先的聊天 AI 語音助手，我們與現有電動車客戶（包括 Lucid、Togg 等）的合作也取得了令人鼓舞的成果。Togg 近期已將業務拓展至整個德國市場。

We've also signed multiple deals with prominent two-wheeler companies based in the expansive Indian market, as well as multinational commercial fleet vehicle company based out of Italy that manufactures light, medium, and heavy vehicles, including trucks, vans, and buses.

我們還與印度龐大市場上的多家知名兩輪車公司以及一家總部位於義大利的跨國商用車輛公司簽署了多項協議，該公司生產輕型、中型和重型車輛，包括卡車、貨車和巴士。

In Financial Services, we continue to work with seven out of the top 10 global financial institutions, with three buying additional services and two signing renewals. We also signed a new enterprise technology deal with a prominent organization supporting credit unions throughout the United States.

在金融服務領域，我們繼續與全球十大金融機構中的七家合作，其中三家購買了額外的服務，兩家簽署了續約協議。我們也與一家為全美各地信用社提供支援的知名機構簽署了一項新的企業技術協議。

In Energy, we signed new contracts with a large utilities company that generates, transmits, distributes, and sells electricity in the United States, and a Texas-based electricity provider serving millions of customers. In region and consumer goods, we had a net new upsell with a major multinational brand with an extensive product portfolio of food, beverage, and consumer goods. And through our smart answering solution, we won deals with one of the fastest-growing global health clubs in the United States and a US-based global franchise that offers state-of-the-art training facilities for elite athletes.

在能源領域，我們與一家在美國從事發電、輸電、配電和售電的大型公用事業公司以及一家為數百萬客戶提供服務的德州電力供應商簽署了新合約。在區域和消費品領域，我們與一家擁有廣泛食品、飲料和消費品產品組合的大型跨國品牌實現了淨新增銷售額。透過我們的智慧應答解決方案，我們贏得了與美國發展最快的全球健身俱樂部之一以及總部位於美國的全球特許經營機構的合作，該機構為精英運動員提供最先進的訓練設施。

In Restaurants, one of our most established verticals, we are a market leader and continue to see strong adoption with our cutting-edge solutions. Notably, we signed a deal to deploy our AI ordering solutions with a nationally recognized full-service restaurant chain and have franchise wins with Firehouse Subs, Five Guys, and McAllister's Deli. And you are now fully rolled out in all MOD Pizza, Habit Burger, Red Lobster, and Torchy's Tacos locations, in addition to existing brands, Chipotle and Casey's.

在餐飲業——我們最成熟的垂直領域之一——我們是市場領導者，我們的尖端解決方案持續獲得強勁的市場認可。值得一提的是，我們與一家全國知名的全方位服務連鎖餐廳簽署了部署人工智慧訂餐解決方案的協議，並贏得了與 Firehouse Subs、Five Guys 和 McAllister's Deli 的特許經營權合作。現在，除了現有的 Chipotle 和 Casey's 品牌之外，您還將在所有 MOD Pizza、Habit Burger、Red Lobster 和 Torchy's Tacos 門市全面推廣。

Pete's Coffee expanded further with Employee Assist, which will now be deployed in all company-owned locations. Earlier this year, we introduced a new product called Voice Insights, targeting brands that require a precursor to full automation, for example, due to missing APIs and infrastructure. Voice Insights is our AI-powered solution that analyzes customer and employee interactions in real time in order to measure efficiency, satisfaction, and other metrics that can help restaurants improve their operations. We are seeing strong interest from our customer brands and prospects, with several brands already in process of rolling it out within just months of being introduced.

Pete's Coffee 透過推出員工協助計畫進一步擴展了服務範圍，該計畫現在將在公司所有自營門市實施。今年早些時候，我們推出了一款名為 Voice Insights 的新產品，目標客戶是那些由於缺乏 API 和基礎設施等原因而需要實現完全自動化的品牌。Voice Insights 是我們基於人工智慧的解決方案，它即時分析顧客和員工的互動，以衡量效率、滿意度和其他指標，從而幫助餐廳改善營運。我們看到客戶品牌和潛在客戶對此表現出濃厚的興趣，一些品牌在產品推出後的短短幾個月內就已經開始推廣應用程式。

In Healthcare, we launched with a large precision medicine provider to pioneer an inbound and outbound agentic AI solution, which is the first of its kind in the healthcare space. We also signed with the US-based regional hospital system to deploy the Amelia platform, and we renewed our relationship with one of the leading healthcare companies for wholesale medical supplies.

在醫療保健領域，我們與一家大型精準醫療提供者合作，率先推出了入站和出站代理 AI 解決方案，這是醫療保健領域首創。我們還與美國一家區域醫院系統簽署了協議，部署 Amelia 平台，並與一家領先的醫療保健公司續簽了醫療用品批發合作關係。

In Insurance, French insurer, Apivia Courtage, announced that it will deploy Amelia 7 to bring agentic AI to its contact centers. We also renewed with a global insurance company that provides services to multinational corporations and a highly regarded Mexico-based insurance company specializing in auto insurance.

在保險領域，法國保險公司 Apivia Courtage 宣布將部署 Amelia 7，為呼叫中心引入智慧代理 AI。我們還與一家為跨國公司提供服務的全球保險公司以及一家備受推崇的墨西哥汽車保險公司續簽了保單。

In Telecommunications, we signed a large, well-known communications provider that offers fiber internet, digital television, and other services to residential and business customers in over 20 US states.

在電信領域，我們與一家大型知名通訊服務供應商簽約，該公司為美國 20 多個州的住宅和商業客戶提供光纖網路、數位電視和其他服務。

In IT services, we renewed a multi-year contract and upsold to one of the largest internet domain registry and web hosting companies in the world. We also won to deal with a leading provider of managed cybersecurity services, cloud and IT infrastructure solutions based out of the United States.

在 IT 服務方面，我們續簽了一份多年合同，並向全球最大的互聯網域名註冊和網絡託管公司之一進行了增值銷售。我們也贏得了與總部位於美國的領先託管網路安全服務、雲端和 IT 基礎設施解決方案供應商的合作機會。

With Channel Partners, we entered into a strategic partnership with leading technology services distributor, Telarus, to bring Amelia 7 and Autonomics to their enterprise CX and EX landscape. Additionally, we signed a multi-year deal with a long-standing partner that specializes in CRM, AI, and workforce engagement management. We entered into a retailer agreement with VOXai, a company that offers purpose-driven customer experience solutions. And entered into strategic partnerships with two of the leading software and service providers of full-suite studio gym, health, and wellness club management.

我們與 Channel Partners 合作，與領先的技術服務經銷商 Telarus 建立了策略合作夥伴關係，將 Amelia 7 和 Autonomics 引入其企業 CX 和 EX 環境。此外，我們還與一家長期合作夥伴簽署了一項多年期協議，該協議專門從事客戶關係管理 (CRM)、人工智慧 (AI) 和員工敬業度管理。我們與 VOXai 公司簽訂了零售商協議，該公司提供以目標為導向的客戶體驗解決方案。並與兩家領先的全套工作室健身房、健康和健身俱樂部管理軟體和服務提供者建立了策略合作夥伴關係。

Many of these important deals and partnerships are a result of our success and growing leadership in enterprise AI. I'm excited to talk more about this increasingly important focus area for SoundHound as we lean in on agentic AI with our differentiating agentic plus framework. With our acquisition of interactions, we've added a number of pre-eminent Fortune 100 companies across various industries to our already strong portfolio of global brands.

這些重要的交易和合作關係，很多都是我們憑藉在企業人工智慧領域的成功和不斷提升的領導地位而達成的。我很高興能進一步探討 SoundHound 在這個日益重要的關注領域，因為我們正憑藉著我們獨特的智慧增強框架，大力發展智慧人工智慧。透過收購互動平台，我們已將許多來自各行業的傑出財富 100 強公司納入我們原本就強大的全球品牌組合中。

For example, we now offer our solutions to one of the largest footwear and apparel brands in the world, to a Silicon Valley-based platform giant, and to some of the major names in automotive, energy, financial services, insurance, healthcare, technology, and telecommunications. Not to mention that we've added hundreds of new patents that we can leverage to increase our innovation modes.

例如，我們現在向全球最大的鞋服品牌之一、矽谷的一家平台巨頭以及汽車、能源、金融服務、保險、醫療保健、科技和電信等行業的一些主要企業提供解決方案。更不用說我們還獲得了數百項新專利，我們可以利用這些專利來增加我們的創新模式。

We see Enterprise AI as one of the biggest near-term opportunities, so we are aggressively expanding our product suite and our customer engagement in that space. On that note, we've just updated the Amelia 7 platform to version 7.3, introducing some capability upgrades that have already been wowing customers and prospects, including major improvements to conversational latency and barging handling to deliver an even more natural and intuitive voice experience for customers.

我們認為企業人工智慧是近期最大的機會之一，因此我們正在積極擴展我們的產品組合併加強在該領域的客戶互動。為此，我們剛剛將 Amelia 7 平台更新到 7.3 版本，引入了一些功能升級，這些升級已經讓客戶和潛在客戶感到驚喜，包括對對話延遲和插話處理的重大改進，從而為客戶提供更加自然和直觀的語音體驗。

The Amelia 7 platform, which offers Enterprise-grade agentic AI, is already making a real impact where it has been deployed. And we are now expanding its availability globally and already seeing brand new logos in our pipeline. Much of this interest is driven by our unique approach to agentic AI, what we call our Agentic+ framework. It's an agentic system designed for enterprises that balances the power of multi-agent orchestration and generative AI with essential business requirements and controls.

Amelia 7 平台提供企業級智慧體 AI，在其部署的地區已經產生了真正的影響。我們現在正在全球擴大其應用範圍，並且已經在籌備推出全新的標誌。這種興趣很大程度上源自於我們獨特的智能體人工智慧方法，我們稱之為 Agentic+ 框架。這是一個專為企業設計的智慧體系統，它將多智能體編排和生成式人工智慧的強大功能與必要的業務需求和控制相結合。

While many internal AI projects stall in pilot phase, our approach consistently brings use cases to market within days or weeks rather than months. Within any given AI agent workflow, there may be certain functions that are more appropriate to complete with predictable deterministic automation, especially when personal security is in question. And there are times when human escalation is necessary or required by our customers.

許多內部人工智慧專案在試點階段停滯不前，而我們的方法卻能在幾天或幾週內（而不是幾個月）將用例推向市場。在任何給定的 AI 代理工作流程中，某些功能可能更適合以可預測的確定性自動化來完成，尤其是在涉及個人安全性時。有時，客戶需要或要求我們採取人工幹預措施。

The complexity and sensitivity of enterprise use cases demand a highly intelligent hybrid agentic system that delivers safe, efficient end-to-end orchestration. Agentic+ provides exactly that, a practical, scalable framework that brings forward-looking AI use cases into real-world operations today. This is where we see the true value unlock in enterprise AI transformation.

企業用例的複雜性和敏感度要求高度智慧的混合代理系統能夠提供安全、高效的端到端編排。Agentic+ 正是為此而生的，它是一個實用、可擴展的框架，可以將前瞻性的 AI 用例引入到當今的現實世界運作中。這就是我們看到企業人工智慧轉型真正價值釋放之處。

As with all of our solutions, this advanced technology has been built upon decades of R&D, troves of data, and an understanding of what motivates our customers. Our platform is LLM agnostic with relevant enterprise integrations, and SoundHound's trademark agility means we will adapt and upgrade Amelia 7 at the pace of AI innovation. We aim to always remain at the cutting edge for our customers.

與我們所有的解決方案一樣，這項先進技術建立在數十年的研發、大量數據以及對客戶動機的理解之上。我們的平台與LLM無關，並具有相關的企業集成，而SoundHound的標誌性敏捷性意味著我們將以人工智慧創新的速度調整和升級Amelia 7。我們的目標是始終為客戶提供最前沿的產品和服務。

That brings me on to Voice Commerce. Voice Commerce is our highly anticipated solution that seamlessly brings conveniences like food ordering and recent additions such as parking payments and restaurant reservations into the vehicle for the convenience of drivers, as well as other IoT devices like TVs. We've now taken POCs to advanced stages with a number of OEMs and merchants, and have already successfully placed live voice orders from cars and completed the transaction. We are looking forward to seeing these go into full production in 2026.

這就引出了語音商務的議題。語音商務是我們備受期待的解決方案，它將訂餐等便利功能以及最近新增的停車繳費和餐廳預訂等功能無縫地引入車內，方便駕駛員使用，還能與其他物聯網設備（如電視）配合使用。我們現在已經與多家汽車製造商和商家將概念驗證推進到了高級階段，並且已經成功地從汽車中發出即時語音訂單並完成了交易。我們期待這些產品在 2026 年全面投入生產。

We have four OEMs showing strong interest in this groundbreaking technology, with others following very closely. One of them in particular is poised to be the first to market together with a large USR. We've also completed the integration with a larger paid parking service provider and a restaurant reservations company. We are on track to have some exciting announcements early next year.

已有四家原始設備製造商對這項突破性技術表現出濃厚的興趣，其他製造商也密切關注。其中一款產品尤其可望率先上市，並附帶一項重要的用戶評估指標。我們也完成了與一家大型付費停車服務提供者和一家餐廳預訂公司的整合。我們預計明年初將發布一些令人振奮的消息。

In addition, Voice Commerce is driving new conversations with smart TV manufacturers. And in particular, we are in talks with two prominent global manufacturers to enable consumers to order food or other services while watching TV simply by speaking to the device. More to come on that in the near future.

此外，語音商務正在推動與智慧電視製造商之間的新對話。尤其值得一提的是，我們正在與兩家全球知名製造商洽談，讓消費者能夠在看電視的同時，透過對著設備說話來訂購食物或其他服務。關於這一點，我們將在近期發布更多相關資訊。

We are nearly fully integrated with two tech platform giants in order to offer our voice ordering to their many millions of users. Going live to consumers is now imminent with more to come on this opportunity. We believe this is a proof point that our decades of relentless innovation is delivering technology that is ready for mass adoption even by big tech. We will have a prominent presence at CES once again in January, showcasing our solutions with participation from several partners. We look forward to seeing some of you there.

我們已與兩家科技平台巨頭實現了近乎完全的整合，以便向其數百萬用戶提供語音訂購服務。即將正式向消費者上線，未來也將推出更多相關產品。我們相信，這證明了我們數十年來不懈的創新正在研發出足以被大規模採用的技術，即使是大型科技公司也會採用。我們將於一月份再次亮相 CES，與多家合作夥伴一起展示我們的解決方案。我們期待在那裡見到你們中的一些人。

In closing, we continue to deliver strong results. Some of the largest companies in the world are coming to us for solutions to address their AI goals. We are at the very beginning of addressing the massive market opportunity in front of us. We are a pioneer in voice and conversational AI, and the expertise we've gained over the past decades are becoming recognized more and more every day. We are delivering value-driven, agentic AI solutions to our customers, and we are ready to offer a voice commerce solution no other company has been able to bring to market.

總之，我們繼續取得了強勁的業績。世界上一些最大的公司正在向我們尋求解決方案，以實現他們的人工智慧目標。我們才剛開始著手解決擺在我們面前的巨大市場機會。我們是語音和對話式人工智慧領域的先驅，我們在過去幾十年中累積的專業知識正日益得到認可。我們正在為客戶提供以價值為導向的智慧AI解決方案，我們準備提供其他公司尚未推向市場的語音商務解決方案。

With that, I'll now turn the call over to Nitesh to talk about our financial performance, key growth drivers, and business outlook.

接下來，我將把電話交給 Nitesh，讓他談談我們的財務表現、關鍵成長驅動因素和業務前景。

Thank you, Keyvan, and good afternoon, everyone. Q3 revenue was $42 million, a 68% year over year. We continue to deliver strong growth led by product and technological differentiation in a rapidly expanding market. Reflecting on our performance so far this year, we have now successfully delivered the Pillar 2 scaling that we had anticipated and communicated last year.

謝謝你，凱文，大家下午好。第三季營收為 4,200 萬美元，年增 68%。在快速擴張的市場中，我們憑藉產品和技術差異化優勢，持續實現強勁成長。回顧今年迄今的表現，我們已經成功實現了去年預期和溝通的第二個支柱規模化目標。

From financial services to healthcare to technology and retail, on top of our existing footprints in automotive and restaurants, we have embedded our leading-edge voice and conversational AI suite deeply into a wide cross-section of market-leading services. The disruptive innovation curve that extends from deep learning and transformer architectures to large language and reasoning models into agentic AI solutions portends societal and economic transformation for decades to come.

從金融服務到醫療保健，再到科技和零售，除了我們在汽車和餐飲領域的現有業務之外，我們還將領先的語音和對話式人工智慧套件深度嵌入到廣泛的市場領先服務中。從深度學習和Transformer架構到大型語言和推理模型，再到智慧AI解決方案，顛覆性創新曲線預示著未來幾十年社會和經濟的變革。

That said, the existing state of AI points vividly to call center and customer service disruption as a current epicenter of this transformation, and our solutions are strategically positioned to capitalize. Our organic and strategic investments have positioned us well to succeed here. From full automation that outperforms humans to human assist capabilities that drive contact center agent efficiencies, we now run the gamut to support enterprises as they deliver best-in-class customer support or outbound lead generation. We have an agentic-first architecture leveraging our own state-of-the-art models alongside best-of-breed partners.

也就是說，人工智慧的現狀清晰地表明，呼叫中心和客戶服務的顛覆是當前這場變革的中心，而我們的解決方案已做好充分的策略準備，可以充分利用這項變革。我們透過自身發展和策略性投資，已經為在這裡取得成功做好了充分準備。從超越人類的全自動化，到提升聯絡中心代理效率的人工輔助功能，我們現在提供全方位的解決方案，以支援企業提供一流的客戶支援或外撥線索開發。我們採用以智能體為先的架構，利用我們自己的最先進模型以及一流的合作夥伴。

With the acquisition of Interactions, we have now added workflow optimization capabilities to our enterprise agentic solutions, stitching the fabric needed to enable companies to effectively adopt AI and deliver productivity and returns. And our deepening broad-based partnerships are a testament that our offerings are resonating.

透過收購 Interactions，我們現在已將工作流程優化功能添加到我們的企業代理解決方案中，從而建立了使公司能夠有效採用人工智慧並提高生產力和回報所需的框架。我們不斷深化廣泛的合作夥伴關係，證明了我們的產品和服務引起了客戶的共鳴。

We have said before, this is the era where natural language conversations will enable humans to more seamlessly interact with technology, and voice AI is the killer app. Our heritage of innovation is our right to win. We continue to see that play out in Q3.

我們之前說過，在這個時代，自然語言對話將使人類更無縫地與科技互動，而語音人工智慧是殺手級應用。我們擁有創新的傳統，這是我們獲勝的權利。我們在第三季繼續看到這種情況發生。

In Pillar 1, we extended our penetration into China with a large IoT win as we capitalized on that country's lead in the global robotics race. In restaurants, another quarter of adding 1,000 locations, including most notably with a leading pizza provider, and expansion beyond ordering with our employee assist and voice insight solutions provided both rapid unit and price expansion.

在第一支柱中，我們利用中國在全球機器人競賽中的領先地位，在物聯網領域取得了重大突破，從而擴大了我們在中國市場的滲透率。在餐飲方面，新增 1000 個地點，其中最值得注意的是與一家領先的披薩供應商的合作，以及透過我們的員工協助和語音洞察解決方案將業務擴展到訂購之外，都帶來了快速的門市數量和價格增長。

In enterprise, our steady retention and expansion rates were supported by significant improvements in customer outcomes. In fact, relative to incumbent solutions, our early agentic AI customers are seeing up to tenfold improvement in containment rates, 25% higher end-user net promoter score, and 15% higher customer satisfaction. And we're achieving these results even faster with up to 35% less effort to design and deploy our agentic AI service. And we are now consistently eclipsing 1 billion queries a month, up nearly 10x since we went public.

在企業客戶方面，我們穩定的客戶留存率和擴張率得益於客戶體驗的顯著提升。事實上，與現有解決方案相比，我們早期的智慧AI客戶在遏制率方面提高了10倍，最終用戶淨推薦值提高了25%，客戶滿意度提高了15%。而且，我們在設計和部署智慧AI服務方面，只需減少高達35%的工作量，就能更快取得這些成果。我們現在每月查詢量穩定超過 10 億次，比上市以來成長了近 10 倍。

Before I move to the quarterly numbers, I want to talk about our pace of investment. The speed of innovation has been rapid the past several years. Now it's about accelerating the adoption curve and customers' realization of AI's massive benefits. The winners will entrench themselves where value can be derived for many years to come. And that's why we are aggressively investing to fortify and expand our moats while deepening our customer relationships.

在介紹季度數據之前，我想先談談我們的投資步伐。過去幾年，創新速度非常迅猛。現在的重點是加快人工智慧的普及速度，讓客戶認識到人工智慧帶來的巨大好處。獲勝者將鞏固自身地位，在未來多年持續創造價值。因此，我們正在積極投資，以鞏固和擴大我們的競爭優勢，同時加深與客戶的關係。

This has been taking the form of go-to-market investments as well as product capability expansion, and we expect to continue to keep the foot on the accelerator. That said, from a financial profile perspective, we are also moving from our past, where our investments were building the future and foreshadowing scale, to our present, where our growth and scale fully covers our costs.

這體現在市場推廣投資和產品能力擴展方面，我們預計將繼續保持這種加速發展動能。也就是說，從財務狀況的角度來看，我們也正在從過去那種投資是為了建立未來、預示規模的模式，轉變為現在那種成長和規模完全覆蓋成本的模式。

More specifically, as we exit 2025 and enter 2026, in part a result of executing on substantial acquisition synergies, we expect continued hypergrowth to be coupled with a break-even profitability profile. I'll share more when we discuss the outlook. For now, let me discuss the third quarter financial results in more detail.

更具體地說，隨著我們告別 2025 年進入 2026 年，部分原因是實現了實質性的收購協同效應，我們預計持續的高速增長將與盈虧平衡的盈利狀況相結合。等我們討論前景時，我會分享更多資訊。現在，讓我更詳細地討論一下第三季的財務表現。

Q3 revenue was $42 million, up 68% year over year. All three pillars grew double digits, and we saw strength in both direct sales and through channel partners. We had a big IoT win for Pillar 1, and enterprise and restaurants helped drive out performance in Pillar 2. While there was continued pressure in the automotive business, driven by global tariffs and the broader industry softness, there are signs of improvement, especially when considering the momentum we are seeing around Pillar 3 voice commerce.

第三季營收為 4,200 萬美元，年增 68%。三大支柱業務均實現了兩大成長，直銷和通路合作夥伴業務均表現強勁。我們在第一支柱領域取得了物聯網方面的巨大成功，而企業和餐飲業則幫助我們在第二支柱領域取得了業績。儘管受全球關稅和整個行業疲軟的影響，汽車行業持續面臨壓力，但也有好轉的跡象，尤其是在考慮到我們所看到的第三支柱語音商務的發展勢頭時。

And as we have substantially diversified our industry mix the past two years, and the individual sector's impact on our growth is much more muted now. Across our business lines, we also expanded geographic reach and product coverage. And we continue to see strong customer diversification where year to date, we don't have any customers contributing greater than 10% of our revenue.

過去兩年，我們大幅實現了產業結構的多元化，因此單一產業對我們成長的影響現在要小得多。我們也擴大了各個業務線的地域覆蓋範圍和產品覆蓋範圍。而且我們持續看到客戶群呈現強勁多元化，今年迄今為止，沒有任何一家客戶的營收貢獻超過我們總營收的 10%。

In Q3, our GAAP and non-GAAP gross margins were both up from the prior quarter. Our GAAP gross margin was 43% and adjusted for non-cash amortization of purchase intangibles and employee stock compensation. Our non-GAAP gross margin was 59%. We continued to drive efficiencies in cloud spend as we deepened our acquisition integration, and we continued to realize cost savings from shifting from third-party solutions to our own home-built ones.

第三季度，我們的GAAP毛利率和非GAAP毛利率都比上一季成長。我們的 GAAP 毛利率為 43%，並已根據購買無形資產的非現金攤銷和員工股票補償進行了調整。我們的非GAAP毛利率為59%。隨著我們不斷深化收購整合，我們在雲端支出方面持續提高效率，並且透過將第三方解決方案轉向我們自己建構的解決方案，我們持續實現了成本節約。

R&D expenses were $22.8 million in Q3, up 17% year over year, largely due to acquisitions and related headcount and data center costs. We continued to invest in innovation to maintain our technological leadership. Our speech foundation model, Polaris, is delivering outstanding results, and we're now deploying it broadly across our customer base. We're also advancing our agentic AI capabilities in real-time speech-to-speech models, leveraging our deep expertise in conversational architectures and machine learning to deliver industry-leading speed and accuracy.

第三季研發支出為 2,280 萬美元，年增 17%，主要原因是收購以及相關的人員和資料中心成本。我們持續加大創新投入，以維持技術領先地位。我們的語音基礎模型 Polaris 取得了卓越的成果，我們現在正在將其廣泛部署到我們的客戶群中。我們也在即時語音對語音模型中推進我們的智慧AI能力，利用我們在對話架構和機器學習方面的深厚專業知識，提供業界領先的速度和準確性。

Sales and marketing expenses were $16.4 million in Q3, reflecting a 96% year-over-year increase, primarily driven by acquisitions. As seen in our results the last few quarters, we have invested heavily in the channel, which is paying dividends. We have also continued to build up direct sales and are also driving demand and lead generation activities while speeding the journey from pipeline to close.

第三季銷售和行銷費用為 1,640 萬美元，年增 96%，主要由收購推動。從我們過去幾季的業績可以看出，我們對該通路進行了大量投資，而這些投資正在獲得回報。我們也持續拓展直接銷售，並積極推動需求和潛在客戶開發活動，同時加速從銷售管道到最終成交的進程。

G&A expenses were $24.3 million in Q3, reflecting a 43% year-over-year increase, primarily driven by our acquisitions. We had roughly $5 million in one-time M&A-related costs. Aside from that, we continue to drive operational efficiencies throughout the organization and improve our control environment. We had non-cash employee stock compensation of $19.7 million and depreciation and amortization, including the amortization of intangibles, of $8.6 million in Q3, all of which are included in our GAAP results.

第三季一般及行政費用為 2,430 萬美元，年增 43%，主要原因是我們的收購活動。我們一次性併購相關費用約為 500 萬美元。除此之外，我們也將持續提高整個組織的營運效率，並改善我們的控制環境。第三季度，我們支付了 1,970 萬美元的非現金員工股票補償金，以及 860 萬美元的折舊和攤提（包括無形資產攤銷），所有這些都計入了我們的 GAAP 結果中。

Adjusted EBITDA was a loss of $14.5 million. OI&E was $7.1 million of income for the quarter. GAAP net loss of $109.3 million and GAAP net loss per share of $0.27 were negatively impacted by the change in fair value of contingent liabilities of approximately $66 million. This relates to the acquisitions we have completed and is a non-operating and non-cash expense and primarily reflects the quarter-on-quarter increase in our stock price. As such, this item has been excluded in our non-GAAP results.

調整後 EBITDA 虧損 1,450 萬美元。 OI&E該季度營收為710萬美元。GAAP淨虧損1.093億美元，GAAP每股淨虧損0.27美元，受到或有負債公允價值變動約6,600萬美元的負面影響。這與我們已完成的收購有關，屬於非經營性非現金支出，主要反映了我們股價的季度環比增長。因此，該項目已從我們的非GAAP績效中剔除。

Non-GAAP net loss was $13 million, and non-GAAP net loss per share was $0.03 in the quarter. This adjusts for items such as non-cash depreciation and amortization, M&A transaction costs, and stock-based compensation. Our balance sheet remains strong, with cash and equivalents at quarter end of $269 million and no debt.

本季非GAAP淨虧損為1,300萬美元，非GAAP每股淨虧損為0.03美元。這調整了非現金折舊和攤提、併購交易成本和股票選擇權等項目。我們的資產負債表依然穩健，截至季末現金及等價物為 2.69 億美元，且無任何債務。

With that, let me discuss our financial outlook. I'll complete a thought I started earlier about where we are in the longer-term trajectory of this business. Keyvan started these prepared remarks by noting our recently celebrated 20th anniversary as a company. He and our founding team started on a path of breakthrough science, tackling the challenge of hard AI, innovation that provides more seamless and natural access so humans can harness the power of technology for our collective benefit, notably through pioneering advances in voice AI.

接下來，我將談談我們的財務前景。我將完成我之前開始的一個想法，即我們在這個行業的長期發展軌跡中所處的位置。Keyvan 在發表事先準備好的演講時，首先提到了我們公司最近慶祝成立 20 週年。他和我們的創始團隊踏上了突破性科學之路，致力於解決硬人工智慧的挑戰，透過創新提供更無縫、更自然的存取方式，使人類能夠利用技術的力量造福於我們所有人，尤其是在語音人工智慧領域取得了開創性進展。

Breakthrough science is challenging. It takes time and requires tenacity and resilience to persist through the cycles of revolution, setbacks, and further evolution. That was our company's existence for the first 15 years, and it manifested in financials that were heavy in R&D spend. The last five years as a company have been about commercialization, product deployment, customer traction, and scale. That set the stage for the acceleration into the high-growth part of the S curve, where we are now.

突破性科學充滿挑戰。要經歷革命、挫折和進一步發展的循環，需要時間、毅力和韌性才能堅持下去。這就是我們公司成立前 15 年的經營模式，其財務表現也反映在研發支出方面。公司過去五年主要致力於商業化、產品部署、客戶拓展和規模化。這為加速進入 S 型曲線的高成長階段奠定了基礎，而我們現在就處於這個階段。

Within this high growth stage, we are crossing the chasm to where we expect our inflows to exceed outflows. That transition, like all others, aren't linear or uniform, but they are progressive and ultimately compounding. That's the setup of our business as we look towards 2026.

在這個高速成長階段，我們正在跨越鴻溝，預計流入資金將超過流出資金。這種轉變，和其他所有轉變一樣，並非線性或均勻的，而是漸進的，最終會產生累積效應。這就是我們展望 2026 年的業務架構。

With that broad context, for the full-year 2025, we now expect revenue to be in the range of $165 million to $180 million. For Q4, we expect to be adjusted EBITDA profitable at the higher end of the revenue outlook and in the single-digit millions of loss at the lower end. We see additional acquisition cost synergies of roughly $20 million on an annual run rate basis to be realized more fully in 2026, which will set us up well as we align our organization with the massive tailwinds behind us.

基於上述大背景，我們現在預計 2025 年全年營收將在 1.65 億至 1.8 億美元之間。對於第四季度，我們預計在收入預期上限的情況下，調整後 EBITDA 將實現盈利；而在收入預期下限的情況下，虧損額將為數百萬美元。我們預計，到 2026 年，每年將有約 2000 萬美元的額外收購成本協同效應得到更充分的實現，這將使我們更好地調整組織架構，以適應我們身後巨大的順風。

Accordingly, our early expectations for 2026 are to continue delivering high growth commensurate with levels we have been compounding the past several years, and we expect to do so with near break-even profitability levels because we want to reinvest when we foresee outsized returns. AI is fundamentally transformative. We have the assets and capabilities to deliver this transformation for our customers, but we will stay aggressive in our approach because we believe the potential value capture merits it, where expected returns are well in excess of the risk-adjusted cost of capital.

因此，我們對 2026 年的初步預期是繼續實現與過去幾年複合增長水平相稱的高增長，並且我們預計盈利水平將接近盈虧平衡，因為我們希望在預見到超額回報時進行再投資。人工智慧從根本上改變了世界。我們擁有為客戶實現此轉型所需的資產和能力，但我們將保持積極進取的態度，因為我們相信潛在的價值獲取值得我們這樣做，預期回報遠遠超過風險調整後的資本成本。

With that, we will now move to Q&A.

接下來，我們將進入問答環節。

(Operator Instructions)

（操作說明）

Gil Luria, DA Davidson.

吉爾·盧裡亞，地方檢察官戴維森。

First, I wanted to ask about the eight-figure Chinese robotics deal. Is it too much for us to think of this as maybe a humanoid robotics? It seems like that may be a good application for low-latency voice-to-meaning, and therefore, a very interesting new development. And then the other part of that question is the double-digit millions over what timeframe?

首先，我想問一下那筆價值數千萬美元的中國機器人交易。我們是否可以將其視為人形機器人？這似乎可以很好地應用於低延遲語音到語義的轉換，因此，這是一個非常有趣的新發展。那麼問題的另一部分是，這數千萬美元是在什麼時間內發生的？

Yeah, it is a robotic company. This particular product is not a human robot, but the deal does pave the way to experiences that you are imagining. This one is more of a devices can carry, and like a wearable, but not actually a wearable, but in that category, and in the double-digit million in the next two to three years.

是的，這是一家機器人公司。這款產品並非人形機器人，但這項交易確實為實現你所想像的體驗鋪平了道路。這款產品更像是可以隨身攜帶的設備，類似於可穿戴設備，但實際上並不是可穿戴設備，而是屬於這一類別，預計在未來兩到三年內銷售將達到數千萬台。

Got it. Hope to be invited to the demo when it goes to Humanoid.

知道了。希望在Humanoid展會上能受邀參加展示。

Yeah, sorry to add that it's actually a commitment number from them. It's not just our SMS.

是的，很抱歉補充一點，這實際上是他們給出的承諾編號。不僅僅是我們的簡訊。

Got it. And then the second one on the interactions acquisition, where do you specifically -- which verticals do you think will specifically have an impact? And then what do you expect the financial impact to be for the balance of the year and into next year?

知道了。其次，關於互動獲取，您認為具體哪些垂直領域會受到具體影響？那麼，您預計這將對今年剩餘時間和明年產生怎樣的財務影響？

Gil, so the first part, there is nice adjacencies and sort of going deeper with our enterprise verticals so they have strength across -- some of their customers overlap in our automotive and our smart tech services. They have a really deep retail footprint and several of the verticals that we're in. And the application of the technology, particularly around the workflow orchestration or they have sort of intent analysts that really complement some of the more complicated enterprise use cases, so you could take all the financial services and healthcare. So there's really nice complement to the existing portfolio.

Gil，所以第一部分，我們與企業垂直領域有很好的關聯，並且正在深入研究，因此他們在各個方面都很有優勢——他們的一些客戶在我們汽車和智慧技術服務領域有重疊。他們在零售領域擁有非常深厚的佈局，並涉足我們所處的幾個垂直行業。該技術的應用，尤其是在工作流程編排方面，或者他們擁有某種意圖分析師，可以真正補充一些更複雜的企業用例，例如金融服務和醫療保健。所以，這確實是一個很好的現有產品組合的補充。

In terms of contribution, I mean, this is, I think it's sort of a pattern of our M&A that's similar to what we've seen where we had brought companies that have amazing customer long-term contracts, customer relationships, trust. Frankly, in some cases, legacy technology that we're able to partner and bring our own innovation on top of and recalibrate the growth curve. So this is one that we're pretty excited about how we can regrow them together.

就貢獻而言，我認為這有點像是我們併購的模式，類似於我們之前收購的公司擁有出色的客戶長期合約、客戶關係和信任。坦白說，在某些情況下，我們可以與傳統技術合作，在其基礎上引入我們自己的創新，並重新調整成長曲線。所以，我們非常興奮地看到我們如何讓他們重新煥發生機。

And so they're all contemplated in our outlook. I think you'll notice a little nudge up in our expectations, particularly with respect to next year. So it's an important acquisition. I think it's one that we're excited about what we're going to do. And most importantly, to your first part, just really excited to compliment that it brings both on a tech product platform as well as sort of industry overlap.

因此，它們都納入了我們的考慮範圍。我想你會注意到我們的預期略有提高，尤其是在明年方面。所以這是一筆重要的收購。我認為這是我們非常期待要做的事情。最重要的是，對於你的第一部分，我非常高興地讚揚它既是一個技術產品平台，又具有某種行業重疊性。

James Fish, Piper Sandler.

詹姆斯·菲什，派珀·桑德勒。

Hi, this is [Caden] on for Fish. I was just wondering, could you provide a percentage of what you're seeing come from term license versus FAS within Amelia at this point? And then anything to call out for a one-time revenue this quarter?

大家好，我是[Caden]，為您帶來Fish的報導。我想問一下，您能否提供一下目前 Amelia 公司內部，長期授權收入與 FAS 收入分別佔多少百分比？那麼，本季是否有任何一次性收入預期呢？

Sure. Hey, Caden. I can give you more of the general trending, and we've continued to grow our recurring footprint with Amelia, and we have noted in prior quarters that there were sort of a greater one-time-type license deals we talked about that last time. Much smaller footprint this quarter.

當然。嘿，凱登。我可以提供更多整體趨勢，我們與 Amelia 的經常性業務持續成長，而且我們在之前的幾個季度也注意到，我們上次討論過一些更大的一次性授權交易。本季度佔地面積大幅縮減。

So I think with respect to Amelia, it continues to be more heavily penetrated towards the recurring, but I think most importantly, as we look at the shift to a and just really what the technology is able to do now integrating LLMs with our deterministic flows, the footprint is you embed, you get a recurring basis, but it's more outcome-based contracts and pricing. So as we can continue to deliver outcomes, for example, in hospitality, we can book more reservations or in maybe a healthcare setting, book more appointments, like the economic model is one that is advantageous that we can scale.

所以我認為，就 Amelia 而言，它繼續更深入地滲透到循環模式中，但我認為最重要的是，當我們審視向 a 的轉變，以及這項技術現在能夠將 LLM 與我們確定性流程相結合的能力時，其影響是，您可以嵌入，您可以獲得循環基礎，但它更側重於基於結果的合約和定價。因此，隨著我們不斷取得成果，例如在酒店業，我們可以預訂更多房間；或者在醫療保健領域，我們可以預約更多服務，這種經濟模式具有優勢，我們可以擴大規模。

So it'll be recurring plus outcome-based generative or incremental revenue. That's sort of the model also with respect to why we comment on restaurants and high order completion rates. Some of our pricing is just fixed per location amounts. And then more and more customers are seeing that there's real sharing of economic upside if we can say it's more based on real returns to the customer and the pricing will follow.

所以它將是經常性收入加上基於結果的生成性或增量性收入。這也是我們對餐廳和高訂單完成率進行評論的原因之一。我們部分商品的價格是按地點固定的。越來越多的客戶意識到，如果我們能夠說價格更多是基於客戶的實際回報，那麼經濟收益就能真正共享，價格也會隨之調整。

Mike Lattimore, Northland Capital Markets.

Mike Lattimore，北地資本市場。

Hi, this is Vijay Devar from Mike Lattimore. How many customers have committed to upgrading to Amelia 7 right now? I think the number was around 15 last quarter.

大家好，我是來自 Mike Lattimore 的 Vijay Devar。目前有多少客戶承諾升級到 Amelia 7？我記得上個季度這個數字大概是15左右。

Yeah, we're continuing to grow. The 15 was sort of a selected first set of customers, and that has progressed really, really well. That was an initial cohort, and they were sort of our early adopter group. That group, that number just continues to grow. We're migrating with others. We've expanded that set quite significantly. We're in active conversations with a number of them.

是的，我們仍在不斷發展壯大。這 15 人算是首批精選客戶，而且進展非常順利。那是最初的一批用戶，他們算是我們的早期採用者群體。這個群體，這個數字還在不斷增加。我們和其他人一起遷移。我們已經大幅擴展了該系列產品。我們正在與他們中的許多人進行積極的對話。

Ultimately, our target here is that about 75% customers, we expect to be moving onto Amelia 7 probably by mid-next year. So we're sort of thinking of the trajectory of moving towards that. And ultimately, all of our new customers are going to get into that -- are going to be migrated onto Amelia 7. So you know, we're trying to make sure there's a fair migration path for all our customers. Obviously, every customer is different, and we need to be thoughtful about their journeys and be very sensitive to their own end customers.

最終，我們的目標是到明年年中，大約 75% 的客戶將升級到 Amelia 7。所以我們正在考慮朝著這個方向發展的方向。最終，我們所有的新客戶都將遷移到 Amelia 7。所以你知道，我們正在努力確保所有客戶都能獲得公平的遷移途徑。顯然，每個客戶都是不同的，我們需要認真考慮他們的體驗歷程，並且非常關注他們的最終客戶。

Ultimately, we're trying to orchestrate across all other platforms, so we make sure that there's interoperability with other agentic platforms, and we're really thoughtful about just onboarding and pace. So continuing to see great momentum, lots of exciting conversations, and most importantly, I've mentioned in my prepared remarks just the outcomes or the feedback we're getting, whether it's in net promoter score, customer satisfaction, or even just -- and the containment rate improvements. Like, they're real positive outcomes early days, so we're trying to be aggressive in how we migrate.

最終，我們試圖協調所有其他平台，因此我們確保與其他代理平台具有互通性，並且我們非常注重用戶引導和進度。因此，我們看到了強勁的發展勢頭，進行了許多激動人心的對話，最重要的是，我在準備好的發言稿中提到了我們收到的成果或回饋，無論是淨推薦值、客戶滿意度，還是控制率的提高。例如，早期階段就取得了非常正面的成果，所以我們正在積極地推動遷移工作。

Well, that's pretty interesting. Second one, on what percent of your revenue is recurring [revenue]?

嗯，這很有意思。第二個問題，您的經常性收入佔總收入的百分比是多少？[收入]？

The vast majority of our revenue is recurring, and there's a different -- well, I guess I'll group it. We talk about recurring and reoccurring. You know, like I've mentioned, I think, A couple times around our Automotive business where we have license recognition as cars are shipped, we get recognition for the voice capabilities and we get a royalty on that. So maybe I'll count that reoccurring as long as these mega OEMs keep shipping cars under the contract duration, we get revenue.

我們的大部分收入都是經常性收入，還有一種不同的——嗯，我想我應該把它歸類。我們來談談循環和重複。你知道，就像我之前提到的，我認為，在我們汽車業務中，有好幾次，我們在汽車出貨時通過許可證識別，我們獲得了語音功能的認可，並從中獲得了版稅。所以，只要這些大型汽車製造商在合約期間繼續交付汽車，我們就能獲得收入，我或許可以把這部分收入算作經常性收入。

Then we have, and to the prior question, from Caden around Amelia. It's recurring, largely. Their SaaS, oftentimes, they're fixed price up to certain levels of interactions. And then if the customer activity grows above an interaction level, then it gets priced to the next level up. And then we do, from time to time, have certain recognition. When we deploy an edge solution where our obligation to the customer is to pass over that license, than their immediate revenue recognition.

然後，對於之前的問題，我們從凱登到阿米莉亞。這種情況經常發生。他們的 SaaS 產品通常採用固定價格，但僅限於一定次數的互動。然後，如果客戶活動超過某個互動級別，則價格將提升到下一個級別。然後，我們偶爾也會獲得某種認可。當我們部署邊緣解決方案時，我們對客戶的義務是移交許可證，而不是立即確認收入。

So the vast majority is recurring, you know, SaaS-like, but there's a diversification in the product suite. And as I mentioned in the other question you asked, more and more we're finding the trends towards outcome-based, and again, the reason is the AI solutions work. They can deliver more value. They can align to the economic interest of the customer, and so it makes sense for us to price accordingly.

所以絕大多數都是循環型的，你知道，類似 SaaS 模式，但產品組合是多樣化的。正如我在你提出的另一個問題中提到的那樣，我們發現越來越多的趨勢是朝著結果導向型發展，而原因再次在於人工智慧解決方案的有效性。它們可以創造更多價值。它們可以與客戶的經濟利益保持一致，因此我們相應地定價是合理的。

Scott Buck, HC Wainwright and Co.

Scott Buck，HC Wainwright and Co.

So you listed off, you know, kind of eight or nine industry verticals in the release and talked through them. Do you feel like you have enough capacity across each of those to continue to grow them? Or if not, how are you kind of prioritizing where your attention goes in your term?

所以你在新聞稿中列出了大約八到九個行業垂直領域，並逐一進行了講解。你覺得你在這些方面都有足夠的能力繼續發展嗎？或者，如果不是這樣，你在任期內是如何安排注意力優先順序的？

Sure, I can start and Keyvan can add. I mean, I guess to think about it, I get this question to be open and we get this question a lot like, are you doing too much or, you know, and I kind of Not trying to be flippant, but if you start hearing us talk about investing in nuclear energy, maybe that's the fair place to say we're extending a little too far.

當然，我可以先開始，Keyvan可以補充。我的意思是，仔細想想，我常被問到這個問題，例如，你們是不是做得太多了？或者，你知道，我不是想輕率地回答，但如果你開始聽到我們談論投資核能，也許可以說我們做得有點過頭了。

We're a horizontal platform. We start with the premise of, like the pioneering vision was in voice AI. We believe that's the major shift that we are going to enable humans to interact with technology predominantly through natural conversations in voice, the way we're talking right now. And we'll be able to get many, many things done. And so we think, first and foremost, in terms of the ultimate vision, that can pervade across many, many industries.

我們是一個橫向平台。我們的出發點是，就像語音人工智慧的開創性願景一樣。我們相信，我們將實現的重大轉變是，使人類能夠主要透過自然的語音對話與科技進行互動，就像我們現在說話的方式一樣。我們將能夠完成很多很多事情。因此，我們認為，首先，從最終願景的角度來看，它可以滲透到許多許多行業。

And again, we've got traction in automotive restaurants moving into health care, financial services, and setting appointments, booking reservations, doing money transfers, all of that's just on the horizontal platform. So when we're deploying our technology, and we've mentioned it previously, whether it's Polaris or some of the other capabilities, we're best of breed. We're market-leading, even against unlimited resource competitors, where we outperform on our speech recognition technology, on how we bring real-time speech to speech for understanding and conversation.

而且，我們在汽車餐飲領域取得了進展，並開始進軍醫療保健、金融服務領域，預約、預訂、轉帳等等，所有這些都只是在橫向平台上進行的。所以，當我們部署我們的技術時（我們之前也提到過），無論是 Polaris 還是其他一些功能，我們都是同類產品中的佼佼者。即使面對擁有無限資源的競爭對手，我們依然處於市場領先地位，我們在語音辨識技術方面表現出色，能夠將即時語音轉化為語音，從而實現理解和對話。

So that, I think, when you think of us as a platform provider, that's sort of like the premise. Again, I'd say that's where our focus is. That's where we determine, as long as we're playing in the game of conversational and voice AI, that's the right focus.

所以，我認為，當你把我們看作一個平台提供者時，這大概就是我們的前提。我再次強調，這就是我們關注的重點。這就是我們認為的重點所在：只要我們還在從事對話式和語音人工智慧領域的工作，這就是正確的焦點。

Now, it is fair, yeah, there are different applications, especially when you go into workflow integrations that, you know, different ecosystems have different, you know, appointment reservation systems or, you know, order-taking or point-of-sale systems in the restaurant. And there are in many of these industries fragmentation. So we do have to be thoughtful about how deep we can go and who we partner with.

現在，這很公平，是的，有不同的應用，尤其是在工作流程整合方面，你知道，不同的生態系統有不同的，例如餐廳的預約預訂系統或點餐或銷售點系統。而且，這些行業中有很多都存在分散化現象。所以我們必須認真考慮我們可以深入到什麼程度，以及與誰合作。

So we've talked in the past about our partnering strategy in restaurants, for example, where some of the drive-through opportunity requires hardware partnerships. And we're excited that we work with the likes of Samsung and HME and PAR and others. So we think that's a great complement. We can go to market together.

所以我們過去曾討論過我們在餐飲業的合作策略，例如，一些得來速服務的機會需要硬體合作。我們很高興能與三星、HME、PAR 等公司合作。所以我們認為這是一個極好的補充。我們可以一起去市場。

With respect to the integration with the menu structures, we're excited that we partner with the likes of Square Toe, Solo, Oracle, MICROS Simphony, and on and on. And so that's an example of where we kind of go, here's where our software extends, and here's where we want to use partners to go deeper. And that same application in some of the announcements we made today with some of the channel partners going deeper into healthcare, that's how we calibrate.

在與菜單結構的整合方面，我們很高興能與 Square Toe、Solo、Oracle、MICROS Simphony 等公司合作。所以，這就是我們努力的方向的一個例子，我們的軟體可以擴展到這裡，而我們希望與合作夥伴一起深入探索這裡。在今天我們與一些通路夥伴宣布的更深入的醫療保健領域中，我們也運用了同樣的策略，這就是我們進行調整的方式。

So I think there's a lot of room. I take the premise out of your point of like, hey, we're hustling relative to the, certainly the big tech, smaller scale company, we have to be very judicious with our limited resources. I actually think that's a strength of ours because it does force prioritization and it forces us to focus on where our main strengths are, but that's definitely something we're constantly calibrating around.

所以我覺得還有很大的發展空間。我從你的觀點中得出這樣的結論：嘿，相對於大型科技公司而言，我們發展速度很快，但規模較小的公司，我們必須非常謹慎地利用有限的資源。我其實認為這是我們的優勢，因為它迫使我們分清輕重緩急，讓我們專注於自身的主要優勢所在，但這絕對是我們一直在不斷調整的事情。

Great. I appreciate the added color there. Second, I'm just curious on the voice commerce launching in '26, laid out in the release, is that something that that you're sharing marketing responsibilities for? Does that fall on the OEMs? I guess, what does that rollout look like?

偉大的。我喜歡那裡增添的色彩。其次，我很好奇新聞稿中提到的將於 2026 年推出的語音商務功能，您是否也負責該功能的行銷工作？這是否應該由原始設備製造商 (OEM) 負責？我想問的是，這個推廣方案會是什麼樣的？

Well, we talked about voice commerce for a number of years. We showcased it at CES of January of this year, in 2025, end-to-end, and it was very well received. Immediately after that, we had multiple OEMs that started running pilots and POCs, and several brands, merchant brands, like national and global brands that are participating. And those are all going really well. It's moving forward. We're getting more traction.

我們已經討論語音商務好幾年了。我們在今年一月的CES展會上完整展示了2025年的產品，並獲得了非常好的迴響。此後不久，就有多家原始設備製造商 (OEM) 開始進行試點和概念驗證，還有幾個品牌、商家品牌，例如國內和國際品牌也參與其中。這些進展都非常順利。事情正在向前發展。我們獲得了更多支持。

But there are some that are actually eager to go live. Some want to be the first to go live. So we feel very confident that it's going to happen. Not everything is in our control because the OEM has to do something. But a lot of the work is being done by us, actually, the whole integration of the voice AI and the merchant experience is done by us.

但也有一些人非常渴望上線。有些人想成為第一個上線的人。所以我們非常有信心這件事一定會發生。並非所有事情都在我們的掌控之中，因為原廠也需要採取一些措施。但實際上，很多工作都是我們自己完成的，語音人工智慧與商家體驗的整個整合都是我們完成的。

We have done an end-to-end. We are able to drive a car, talk to the car, place an order, go pick it up from the store. All of that is done. We'll have more to show at CES and hopefully, more to share about the timing of an actual go live in production around that time.

我們已經完成了整個流程。我們可以開車，跟車子說話，下單，然後去商店取貨。所有這些都完成了。我們將在 CES 上展示更多內容，並希望屆時能分享更多關於實際上線生產環境時間的資訊。

Leo Carpio, Joseph Gunner.

利奧·卡皮奧，約瑟夫·岡納。

A couple of quick questions. First, on the competitive environment. Can you give us an update on the competitive environment? Are you still facing off against the vendors that we've talked about in the past? And how have the large LLMs competing space. I mean, that they started to encroach. But ultimately, how deep is your competitive moat?

幾個簡單的問題。首先，我們來談談競爭環境。能否介紹一下當前的競爭環境？你還在和我們之前討論過的那些供應商競爭嗎？大型LLM專案在競爭空間中處於什麼地位？我的意思是，他們開始侵占了。但歸根結底，你的競爭護城河有多深？

And then turning, secondly, onto the contracts that you've won, have you been seeing any pricing pressure at all, or it's pretty much you're getting the pricing that you asked for at this stage of the adoption curve?

其次，就您贏得的合約而言，您是否感受到任何價格壓力，或者說，在目前的採用曲線階段，您基本上都能獲得您要求的價格？

Yeah, so I'll talk about the competition. First of all, the space is extremely attractive and you hear more names and that's more of a validation. And we've had competitors in our whole life of 20 years and had bigger competitors in the past. The particular space we're going after, enterprise AI, customer service, we feel we are the leader because of the 20 years of innovation. We have our own technology.

好的，那我來談談比賽吧。首先，這個領域非常有吸引力，你會聽到更多人的名字，這本身就是一種認可。在過去的20年裡，我們一直都有競爭對手，過去也遇過規模更大的競爭對手。我們所瞄準的特定領域，即企業人工智慧和客戶服務，我們認為我們是該領域的領導者，因為我們擁有 20 年的創新經驗。我們擁有自己的技術。

Most of the new players don't have their own technology, so they're using APIs and models from third-party, and they have to kind of stick it together and make it work, and a lot of these models make really good POCs and good demos, but when you go to production, they have issues. But we are able to actually go from demo to deployment in production faster with a higher quality. And that's thanks to our 20 years of experience, having our own models, lower cost, higher accuracy, better latency, more integrations that we've accumulated over the years and some through the acquisition.

大多數新玩家沒有自己的技術，所以他們使用第三方 API 和模型，他們必須把它們拼湊起來才能運行，許多這樣的模型可以作為非常好的概念驗證和演示，但當投入生產時，它們就會出現問題。但我們實際上能夠以更高的品質更快地將演示版本部署到生產環境中。這得益於我們 20 年的經驗、我們自己的模型、更低的成本、更高的準確性、更好的延遲、更多集成，這些都是我們多年來積累的，有些是透過收購獲得的。

So we feel very confident about the space. And some of the names that you may hear in the market could end up being our customers because they need models from companies like SoundHound. And as they go through their choices, they will learn that models like Polaris outperform the competition, right? They beat the big tech and some of the industry giants by, you know, as much as 35%, 40% in accuracy, several times in latency, I mean, at a lower cost.

所以我們對這個空間非常有信心。你在市場上可能會聽到的一些名字最終可能會成為我們的客戶，因為他們需要像 SoundHound 這樣的公司的產品。當他們做出選擇時，他們會發現像北極星這樣的車型性能優於競爭對手，對嗎？他們的準確率比大型科技公司和一些行業巨頭高出 35% 到 40%，延遲低好幾倍，而且成本更低。

I think your second question was around pricing. You know, I'm going to make the general point, and Keyvan can certainly add color. I think in a lot of these sort of eras where we're shifting from old tech to new tech, people are really sort of there's pressure on if you provide legacy technology to drop prices, and it becomes a little bit of a price battle. And so the key is, we're demonstrating our innovation is where you can showcase price value, and alignment, and that's where you can protect on pricing.

我認為你的第二個問題是關於價格的。你知道，我只是想表達一個大致的觀點，而 Keyvan 當然可以增添一些色彩。我認為在許多這樣的時代，當我們從舊技術過渡到新技術時，如果你提供傳統技術，就會面臨很大的壓力，迫使你降低價格，這就會演變成一場價格戰。因此，關鍵在於，我們要展示我們的創新之處在於，您可以展示價格價值和一致性，而這正是您可以在價格方面進行保護的地方。

And we've shown that and said that, I think, in prepared remarks today and previously around where we are seeing pricing expansion. A lot of that is because we're bringing innovation. We're layering on top generative AI intersection. The first company to bring generative AI into the vehicle at Stellantis early last year.

我們已經展示並表達了這一點，我認為，在今天的發言稿中以及之前關於價格擴張的發言中都提到過這一點。這很大程度上是因為我們不斷推行創新。我們正在疊加生成式人工智慧交叉技術。去年年初，Stellantis成為首家將生成式人工智慧引入汽車領域的公司。

We're seeing that with agentic capabilities. Like, you need to be competitive in an ecosystem, as Keyvan noted, that's increasingly competitive. But the use case opportunities and expansions are so tremendous. So if you're a hospitality and you believe through a nice conversational engine that it delights the customer, you can actually upsell a reservation or, you know, provide more services, then you're willing to pay more. And that's a little bit the transition we're in.

我們從智能體能力中看到了這一點。正如 Keyvan 指出的那樣，你需要在一個競爭日益激烈的生態系統中保持競爭力。但是，其應用場景和擴展潛力非常巨大。所以，如果你從事酒店業，並且相信透過良好的溝通技巧可以取悅顧客，從而促進預訂升級或提供更多服務，那麼你就願意支付更多費用。這就是我們目前所處的轉型階段。

So I don't want to disparage, you know, in certain parts, certainly because we're across industry, there's different stories for different sectors and different solutions. And there's macroeconomic dynamics that play into it. But I just say largely there's a transition of pricing architectures. I think ultimately for those who can provide real innovation and product quality, there's ultimate ASP expansion available. And so there's a little bit of both sides of that equation from a pricing dynamic that we're navigating through right now.

所以我不想貶低某些方面，你知道，當然，因為我們涉及各個行業，不同的行業有不同的故事和不同的解決方案。而且，宏觀經濟動態也對此產生影響。但我認為，定價架構總體上正在轉變。我認為，最終對於那些能夠提供真正創新和產品品質的人來說，ASP（平均售價）的擴張空間是巨大的。因此，我們目前正在應對的價格動態中，既包含等式的正反兩面，也包含正反兩面。

Thank you very much.

非常感謝。

Thank you. This concludes the question and answer session. Thank you for your participation in today's conference. This does conclude the program, and you may now disconnect.

謝謝。問答環節到此結束。感謝您參加今天的會議。程式到此結束，您可以斷開連線了。