{"url_path":"/sec/aixi/10-k/2026/item-4","section_key":"item-4","section_title":"Item 4 Information on the Company.","topic":"sec","document":{"doc_type":"20-F","doc_date":"2026-05-15","source_url":"https://www.sec.gov/Archives/edgar/data/1935172/0001213900-26-057986-index.html","accession_number":"0001213900-26-057986","cik":"0001935172","ticker":"AIXI","issuer_name":"Xiao-I Corp","edgar_url":"https://www.sec.gov/Archives/edgar/data/1935172/0001213900-26-057986-index.html","primary_entity_key":"0001935172","primary_entity_name":"Xiao-I Corp"},"word_count":9205,"has_tables":true,"body_markdown":"Item 4. Information on the Company.\n\n \n\nA. History and Development of the\nCompany.\n\n \n\nIn the following discussion of corporate history,\n“we,” “us,” or “our” refer to Xiao-I.\n\n \n\nHistory of Xiao-I Corporation \n\n \n\nWe were incorporated in the Cayman Islands on\nAugust 13, 2018, with limited liability under the Companies Act. Upon incorporation, the authorized share capital of our Company was US$50,000\ndivided into 1,000,000,000 shares, par value of US$0.00005 each, comprising of 1,000,000,000 Ordinary Shares of a par value of US$0.00005\neach.\n\n \n\nOn August 30, 2018, we established our wholly\nowned subsidiary AI Plus Holding Limited (“AI Plus”), under the law of British Virgin Islands, as our intermediate holding\ncompany, which then established its wholly owned subsidiary, Xiao-I Technology Limited (“Xiao-I Technology”) under the law\nof Hong Kong, which in turn established a wholly owned PRC subsidiary, Zhizhen Artificial Technology (Shanghai) Company Limited (“Zhizhen\nTechnology”) or the WFOE, on March 29, 2019. Subsequently, we, through our WFOE, entered into a series of contractual arrangements\nwith Shanghai Xiao-I and its shareholders whereby we were established as the primary beneficiary of Shanghai Xiao-I for accounting purposes.\nWe have recognized the net assets of Shanghai Xiao-I at historical cost with no change in basis in the consolidated financial statements\nupon the completion of this reorganization. On May 18, 2023, AI Plus established its wholly owned US subsidiary, Xiao-I Plus Inc. On\nMarch 1, 2024, AI Plus established its wholly owned United Arab Emirates subsidiary, Xiao-I Super Ltd. On September 24, 2024, AJ PLUS\nINVESTMENTS INC invested $667,000 in Xiao-I Plus Inc., acquiring a 40% ownership stake, with AI Plus retaining 60%.\n\n \n\nAs of the date of this annual report, Al Plus,\nXiao-I Technology, Zhizhen Technology and Xiao-I Super Ltd. do not have any substantive business operations. As a result of our indirect\nownership in Zhizhen Technology and the variable interest entity contractual arrangements, we are regarded as the primary beneficiary\nof the VIE for accounting purposes. We treat the PRC operating entities as our consolidated affiliated entities under U.S. GAAP, and\nhave consolidated the financial results of these entities in our consolidated financial statements in accordance with U.S. GAAP.\n\n \n\nOn December 13, 2023, we issued 3,700,000 preferred\nshares, each with a par value of US$0.00005 and carrying a voting right equivalent to 20 votes (the “3.7 million Preferred Shares”\nor the “Preferred Shares”) to ZunTian Holding Limited (“ZunTian”), an existing shareholder of Xiao-I (the “Issuance”).\nZunTian is a BVI-incorporated company wholly owned and controlled by Mr. Hui Yuan (“Mr. Yuan”). Mr. Yuan is the CEO and Chairman\nof the Company and a recognized A1 industry key opinion leader and domain expert. As of the date of this annual report Mr. Yuan beneficially\nowns more than 50% of the voting power of Xiao-I. Under the Nasdaq listing rules, the Issuance resulted in a change in control and the\nCompany became a “controlled company” as defined under those rules. As a “controlled company,” we are permitted\nto elect not to comply with certain corporate governance requirements. If we rely on these exemptions, you will not have the same protection\nafforded to shareholders of companies that are subject to these corporate governance requirements.\n\n \n\nRecent Developments\n\n** **\n\nCapital Raising Activities\n\n \n\nOn June 17, 2024, Xiao-I Corporation entered into a Securities Purchase\nAgreement with an institutional investor for the sale of senior convertible notes in the aggregate principal amount of approximately $3.26\nmillion, issued at an 8% original issue discount. The notes are convertible into ADSs (each representing one-third of an ordinary share\nat the time). Concurrently, the company sold 1,000,002 ADSs at par value. The offering was made pursuant to Xiao-I’s effective Form\nF-3 shelf registration statement and a prospectus supplement filed on the same date. The notes bear interest at 6% per annum, are convertible\nat a fixed price, and are secured by a pledge over certain assets. Additionally, the agreement provides the investor with registration\nrights, anti-dilution protection, and customary covenants including restrictions on further debt issuance without consent. Gross proceeds\nfrom the transaction were approximately $2.99 million.\n\n \n\nOn October 30, 2024, Xiao-I entered into a Securities Purchase Agreement\nwith the same institutional investor for the issuance of a $2,175,000 convertible promissory note, sold with an 8% original issue discount.\nThe note is convertible into ADSs representing the company’s ordinary shares at a fixed conversion price and matures in one year,\nwith an annual interest rate of 6%. In connection with the note issuance, Xiao-I issued 550,000 ADSs at par value. Proceeds from the transaction\nwere to be used for general corporate purposes. The note includes standard covenants and restrictions, including a limitation on additional\nindebtedness and customary registration rights for the underlying securities.\n\n \n\n52\n\n \n\n \n\nOn January 6, 2025, Xiao-I entered into two separate Securities Purchase\nAgreements with two institutional investors, issuing convertible promissory notes in an aggregate principal amount of $4,637,840, composed\nof an initial principal of $4,295,000 and an additional $342,840 of original issue discount to be added 30 days after issuance. The notes\nbear interest at 6% per annum and are convertible into ADSs at a conversion price equal to 85% of the lowest daily VWAP (volume-weighted\naverage price) during the 10 consecutive trading days prior to the conversion date. The company also agreed to file a resale registration\nstatement for the conversion shares. These notes contain customary anti-dilution provisions, adjustment rights upon stock splits and similar\ncorporate events, and restrictions against future financings on more favorable terms (most-favored-nation provisions). Gross proceeds\nto the company were $4.295 million.\n\n \n\nThe January 6 transaction was conducted through\na registered direct offering under the existing shelf registration, and a related Form 424B5 prospectus supplement was filed. The company\ngranted the investors standard covenants, including restrictions on additional indebtedness and a limitation on further equity issuances\nfor a specified period.\n\n \n\nLegal Proceedings\n\n \n\nXiao-I is currently involved in multiple legal proceedings and disputes,\nincluding securities class actions, intellectual property litigation, and labor-related claims, please refer to “*Item 8. Financial\nInformation – A. Consolidated Statements and Other Financial Information – Litigation*” for more details and the\nrecent developments of these proceedings and disputes.\n\n \n\nCorporate Actions and Governance\n\n  \n\nOn August 23, 2024, Xiao-I Corporation implemented a change in its\nADS ratio from one ADS representing one-third of an ordinary share to one ADS representing three ordinary shares, effectively constituting\na one-for-nine reverse ADS split. This move aimed to boost the ADS trading price and regain compliance with Nasdaq’s minimum bid\nprice requirement.\n\n \n\nPreviously, on July 11, 2024, the company received\na notification letter from the Nasdaq Stock Market, indicating that it was not in compliance with Nasdaq Listing Rule 5450(a)(1) because\nthe closing bid price of its ADSs had been below $1.00 for 30 consecutive business days. This deficiency notice was disclosed in a Form\n6-K filed on July 15, 2024. The company was given 180 calendar days, until January 7, 2025, to regain compliance.\n\n \n\nFollowing the ADS ratio change, on September 9,\n2024, Nasdaq notified Xiao-I that it had regained compliance with Listing Rule 5450(a)(1), having maintained a closing bid price of at\nleast $1.00 per ADS for ten consecutive business days.\n\n \n\nB. Business Overview.\n\n \n\n**In the following discussion of business, “we,”\n“us,” or “our” refer to Shanghai Xiao-I and its subsidiaries.**\n\n \n\nOverview\n\n \n\nXiao-I is a holding company incorporated in Cayman\nIslands. As a holding company with no material operation of its own, it conducts substantially all our operations in China through a variable\ninterest entity, or the VIE, Shanghai Xiao-I Robot Technology Co., Ltd., (“Shanghai Xiao-I”) and its subsidiaries.\n\n \n\nShanghai Yingsi Software Technology Co., Ltd.\n(“Incesoft”) was founded in 2001. Incesoft established the Xiaoi robot brand (Chinese: 小i机器人)\nand developed AI technology used to support its consumer-to-consumer business model. In 2009, Incesoft transformed its business model\nfrom consumer-to-consumer to business-to-business. At the same time, founders of Incesoft founded Shanghai Xiao-I, the VIE, which acquired\nthe Xiaoi robot brand and Incesoft’s core AI technology. Following the acquisition, Incesoft was dissolved by de-registering with\nlocal company registrar in accordance with PRC law in 2012. Since 2009, Shanghai Xiao-I has become a leading artificial intelligence (“AI”)\ncompany by building on its wide technology commercialization, brand recognition and culture of innovation in China.\n\n \n\n53\n\n \n\n \n\n**Milestone Accomplishments over 20 Years History**\n\n \n\n \n\nWe are a global leading cognitive artificial intelligence\ncompany. Since our establishment in 2001, We have been dedicated to continuous innovation and breakthroughs in core technologies related\nto cognitive intelligence rooted in natural language processing. our development goal is to achieve scalable implementation and commercialization\nof our innovative proprietary technologies.\n\n \n\nWe, with over 22 years of technical accumulation\nand industry experiences, have become a leading force in the field of AI industrial application. The company adheres to the mission of\n“serve and benefit more people with our AI technology” and focuses on the continuous innovation and breakthrough of artificial\nintelligence technology development.\n\n \n\nWe believe that we are the pioneer of virtual\nchatbot technology. We launched our first chatbot in 2004. Within two years, we applied chatbot technology to the field of intelligent\nservices and took the lead in creating industry application benchmark cases. We developed thousands of business cases, and provide our\ncustomers with a wide range of solutions from diversified products to superior customized services, formulating a scale of business applications\nand a mature commercialization path, establishing our leading position in the artificial intelligence industry.\n\n \n\nAs a representative enterprise in the field of\nCognitive AI, we led the development of the world’s first international standard in affective computing, contributed to the drafting\nof the “China Artificial Intelligence Industry Intellectual Property White Paper” for four consecutive years (from 2010 to\n2013), demonstrated its influence in global artificial intelligence industry. As of May 1, 2026, Xiao-I have 368 authorized patents,\nalong with 146 pieces of software copyrights, 246 registered trademarks, and its accumulation of intellectual property demonstrates the\ncompany’s fruitful achievements in technological innovation. We are also regarded by Gartner as the “representative of Conversational\nAl enterprises”, proving the company’s outstanding position and influence in the industry worldwide.\n\n \n\n54\n\n \n\n \n\nOn June 29, 2023, we launched “Our Own ChatGPT” –\nHua Zang Universal Large Language Model, which has a comprehensive coverage of hundreds of capabilities of the LLM. It possesses the core\nfeatures of “Controllable, Customizable, and Deliverable”, solving the key challenges faced by global AI models. Building\non the solid foundation of the Hua Zang Universal Large Language Model, we launched the revolutionary Hua Zang Ecosystem on October 26,\n2023 which carries significant implications in the industry. Based on the Hua Zang Universal Large Language Model, through Hua Zang Developer\nPlatform, it provides service guarantees such as cultivation, marketing, and investment, connecting global ecosystem partners, customers,\nand developers. It constantly promotes the implementation and commercialization of the customer application scenarios and use cases by\nHua Zang LLM.\n\n \n\nCurrently, dozens of co-created achievements have\nbeen successfully implemented, helping partners further commercialize in vertical fields. Hua Zang Ecosystem is now continuously collaborating\nwith thousands of ecosystem partners, covering 50+ industry fields, involving various fields such as IoT, finance, healthcare, maternal\nand infant, automobile, manufacturing, operator etc. This validates the commercialization path of the Hua Zang Ecosystem and lays a solid\nfoundation for further promoting the large-scale implementation of Hua Zang Universal Large Language Model in various industries.\n\n \n\nActively expanding into the international markets\nis a key driver of future revenue growth for Xiao-I. We set up our APAC headquarter in Hong Kong in 2018. In 2023, Xiao-I became publicly\ntraded on the Nasdaq Stock Market (NASDAQ: AIXI). In 2024, we established Xiao-I Super Limited, a wholly owned subsidiary in the United\nArab Emirates, and XIAO-I PLUS INC., a United States holding subsidiary in which we hold a 60% equity interest, to support and implement\nour global business expansion strategy. \n\n \n\nProduct and Technology Overview\n\n \n\n**Overall Architecture of Xiao-I Products and\nTechnologies**\n\n \n\n \n\n55\n\n \n\n \n\n**Prior to the introduction of Hua Zang LLM\nin June 2023, the overall architecture of our products and technologies is divided into three layers: (1) infrastructure, (2) aggregation\nempowerment platform and (3) domain application. With the participation of Hua Zang LLM, we reformed the product line into Model as a\nservice (MaaS) and non-MaaS.**\n\n \n\n**Infrastructure Layer**\n\n \n\nOur infrastructure layer provides the informational\nsupport for our products and technologies. Typically built with third-party products and technologies, we integrate the information into\nthe infrastructure layer. Additional properties include:\n\n \n\n●Compatibility\nwith cloud native and private or third-party cloud platforms;\n\n \n\n●Ubiquitous\nperception layer connection enabling integration with the Internet of Things, the Internet, 5G, and dedicated networks; and\n\n \n\n●Multidimensional\ndata collection and integration, including spatiotemporal, channels, and community.\n\n \n\n**Aggregation Empowerment Platform Layer**\n\n \n\n**AI Core Technology Platform — Cognitive\nIntelligence Artificial Intelligence (CIAI)**\n\n \n\n \n\nUsing proprietary intellectual property technologies,\nwe have independently developed CIAI, our core technology platform. To date, we have developed and commercialized six core technologies\nbased on CIAI: (1) natural language processing, (2) speech processing, (3) computer vision, (4) machine learning, (5) affective computing\nand (6) data intelligence and hyperautomation.\n\n \n\n56\n\n \n\n \n\n●Natural\nLanguage Processing\n\n \n\n●CIAI’s\nmultilingual, natural language processing capability extracts and analyzes information, mines text, constructs knowledge, and performs\nknowledge representation and reasoning based on words, phrases, sentences, and text, providing solutions to the human-computer interaction\nneeds of diverse enterprises and professional users.\n\n \n\n●Speech\nProcessing\n\n \n\n●The\nhybrid architecture of Time-Delay Neural Network + Deep Feedforward Sequential Memory Network + attention, in combination with our vast\ncorpus accumulation of more than ten years, has enabled us to train our intelligent voice technology for end-to-end application across\nvarious scenarios in numerous fields. Based on these technologies, we have built a variety of intelligent voice solutions under the Aviation\nIndustry Computer-Based Training Committee framework, including intelligent Interactive Voice Response navigation, intelligent outbound\ncall, intelligent agent assistance, intelligent voice quality inspection, and intelligent coaching.\n\n \n\n●Computer\nVision\n\n \n\n●We\noffer various computer vision capabilities, including face recognition and analysis, multi-target tracking, human posture and action\nrecognition, and scene analysis capabilities such as semantic and instance segmentation. In terms of Optical Character Recognition (“OCR”),\nwe have general OCR and customized OCR for all types of cards, invoice, receipts, tickets, and more. In terms of construction drawing\nanalysis, we apply various capabilities including pattern recognition and computer vision to comprehensively analyze and process CAD\ndrawings, bringing to life standard review capability for construction drawings. Relating to engineering, we provide rapid engineering\ncustomization through its internally-developed deep learning framework. We also offer model distillation and pruning solutions to meet\nclients’ model compression requirements. This high performance framework is adaptable to various environments.\n\n \n\n●Machine\nLearning\n\n \n\n●Machine\nlearning methods offered by us include everything from traditional machine learning to the latest deep learning, reinforcement learning,\nactive learning, transfer learning, and generative adversarial networks (“GAN”). These methods are applied across multiple\nfields such as natural language processing, speech recognition, vision recognition and analysis, and in business scenarios such as precision\nmarketing, personalized recommendation, and risk assessment in combination with massive data and distribution processing algorithms to\nform an efficient human-computer collaborative learning system.\n\n \n\n●Affective\nComputing\n\n \n\n●Deep\nlearning technology is used to recognize, understand, process, and simulate human emotions, so as to realize multi-dimensional and multimodal\naffective computing capabilities such as text, voice and vision. We have built affective computing, analysis, and interactive processing\ncapabilities that process real-time perception, intelligent planning, automatic simulation, and this technology has been widely used\nin various practical business scenarios.\n\n \n\n●Data\nIntelligence and Hyperautomation\n\n \n\n●Large-scale\nmachine learning technology mines, analyzes, and processes massive amounts of data, the assets of which are comprehensively integrated\nto extract information contained therein. Business processes are automatically and quickly identified, reviewed, and executed in combination\nwith innovative technologies such as process automation and low code. The results enable enterprises to delegate simple tasks with high\nrepeatability, as well as complex tasks, to AI and data enhancement, thereby improving the quality and efficiency of business operations.\nApplications include data monitoring, data analysis, user profiling, business process automation, financing business automation, financial\nbusiness automation, supply chain business automation, IT operation, and maintenance and integration automation.\n\n \n\n57\n\n \n\n \n\n**Our Product Platforms**\n\n \n\nWe have commercialized our six core technologies\nto create the following product platforms: (1) Conversational AI, (2) Knowledge Fusion, (3) Intelligence Voice, (4) Hyperautomation, (5)\nData Intelligence, (6) Cloud, (7) Intelligent Construction Support, (8) Vision Analysis, (9) Intelligent Hardware Support, and (10) Metaverse.\n\n \n\n●Conversational\nAI Platform\n\n \n\n●Our\nconversational AI platform makes full use of deep learning, data enhancement, and active learning technologies, employing flexible and\ndiverse dialog management and context processing mechanisms, and driven by a powerful learning system, the results of which achieve in-depth\nscenario dialog processing, intent recognition, and complex logic reasoning in combination with structured knowledge and semantic analysis\ncapabilities. Additionally, the platform realizes the business value of conversational AI in a variety of application scenarios, including\nintelligent customer service, smart marketing, intelligent hardware, intelligent assistant, agent assistance, and intelligent human-computer\ntraining.\n\n \n\n●Knowledge\nFusion Platform\n\n \n\n●The\nknowledge fusion platform integrates various types of knowledge such as Q&A, documents, multimedia, information forms, business processes,\nknowledge graphs, and multimodal to assist enterprises in improving knowledge management capabilities, building intelligent service cores,\nsupporting intelligent knowledge management, retrieval, recommendation, application assistance, cognitive reasoning, and other capabilities.\nIt helps enterprise-level intelligent applications, improves work efficiency, optimizes user experience, and reduces enterprise operating\ncosts.\n\n \n\n●Intelligent\nVoice Platform\n\n \n\n●Our\nintelligent voice platform (“IVP”) uses natural language processing (“NLP”), automatic speech recognition, voiceprint\nrecognition, and text-to-speech technologies with human-computer interaction as its core, in combination with various business scenarios,\nto comprehensively create or enhance business capabilities such as intelligent speech solutions, thereby realizing the macro processes\nof intelligent IVP, intelligent outbound calls, speech analysis, agent assistance, and human-computer interaction.\n\n \n\n●Hyperautomation\nPlatform\n\n \n\n●The\nhyperautomation platform innovatively uses low code technology in combination with agents to realize and expand vast capabilities of\nthe traditional low code platform and Robotic Process Automation. It integrates technologies such as OCR, NLP, and visualized data mining\nand analysis, enables users to realize business and process automation, combines capabilities of knowledge base and imitation learning,\nand enables realization of business and process intelligence with intelligent planning capabilities.\n\n \n\n●Data\nIntelligence Platform\n\n \n\n●The\ndata intelligence platform comprehensively integrates data assets, manages the entire life cycle of data, realizes the entire cycles\nof data integration, processing, transformation, analysis, and mining through What You See Is What You Get with the support of component-based\ndata visualization technology. It also helps clients extract valuable information contained in data, and provides assistance in business\nand process automation, business prediction, decision support, among others, and improves the efficiency of data-driven business intelligence\nand business intelligence services.\n\n \n\n●Cloud\nPlatform\n\n \n\n●The\ncloud platform is a comprehensive platform that integrates our various core technical capabilities, such as NLP capability, speech recognition\ncapability, image recognition capability, data analysis capability, etc. The platform can provide fast and simple access to different\ntechnical capabilities for various customers and can also provide independent technical capabilities for customers of different types\nand even industries. Enterprises can flexibly configure according to the technical capabilities of the platform. The platform has features\nsuch as rich capabilities, simple and easy to use, flexible structure, and strong scalability. Whether it is improving customer service\nlevel, increasing service types and content, or expanding technical capabilities, the platform can easily expand and support.\n\n \n\n58\n\n \n\n \n\n●Intelligent\nConstruction Support Platform\n\n \n\n●Our\nintelligent construction support platform offers many capabilities such as parsing, reconstruction, visualization, and multi-dimensional\nanalysis of construction drawings. Combined with a variety of construction application scenarios, the platform can realize intelligent\nconstruction drawings review, design assistance, online collaborative design, among other applications. It enables the construction industry\nto reduce the cost of drawing review, improve per-capita energy efficiency, empowers the construction industry value chain, and facilitates\nthe transformation and upgrading of intelligence and automation.\n\n \n\n●Vision\nAnalysis Platform\n\n \n\n●The\nvision analysis platform uses a variety of computer vision-related technologies to apply OCR, detection, video, and image analysis, helps\nclients extract and mine valuable information contained in images, and realizes business automation, industrial defect detection, monitoring\nanalysis, and other innovative applications encountered in specific business scenarios.\n\n \n\n●Intelligent\nHardware Support Platform\n\n \n\n●The\nintelligent hardware support platform provides the framework of signal collection, processing, analysis, prediction, and more. This framework\ncan be combined with various sensors to quickly process signal, select and adapt appropriate machine learning algorithms for business\nmodeling according to the intelligent requirements of various types of hardware, make full use of various machine learning capabilities\nto make the equipment be more intelligent.\n\n \n\n●Metaverse\nPlatform\n\n \n\n●We\ndeveloped the first virtual digital human in 2016 and released it for the first time at the Guiyang Digital Expo in 2017. We continue\nto innovate and develop more advanced and smarter digital human products. Digital human with multimodal emotional interaction capabilities\ncan be widely used in various business scenarios including film and television production, media, games, financial services, culture,\ntourism, education, healthcare, and retail.\n\n \n\n**Domain Application Layer**\n\n \n\nFor more than 20 years, we have applied our aggregation\nplatform to form a number of mature application fields designed to address the business needs of various fields, including (1) AI + Contact\nCenter, (2) AI + Finance, (3) AI + Urban Public Service, (4) AI + Construction, (5) AI + Metaverse, (6) AI + Manufacturing and (7) AI\n+ Smart Healthcare.\n\n \n\nOur technologies are based, in significant part,\nupon our proprietary intellectual property portfolio. As of May 1, 2026, we have applied for 662 patents, 368 of which have been granted\nand we have obtained 246 registered trademarks and 146 computer software copyrights. In June 2020, the Company passed the national intellectual\nproperty management system certification and obtained the certificate. This certificate represents that the company’s intellectual\nproperty management system conforms to the GB/T 29490-2013 standard. We continue to develop and improve our intellectual property portfolio\nthrough our deep R&D department. As of May 1, 2026, our restructured team consisted of 58 total employees, including 27 research and\ndevelopment specialists, representing 46.55% of our workforce with 11.11% of them holding advanced degrees.\n\n \n\n59\n\n \n\n \n\nOur primary services are software services provided\nby our cloud platform. Software services refer to the sales of software products corresponding to the Company’s obtained patents\nor software copyrights to customers for meeting the needs of different customers in different industries for artificial intelligence:\n\n \n\n(1) *Contact Center*:\nWe leverage contact center AI solutions to improve customer experience and operational efficiency. We offer AI-based platforms, software\ntools and services that leverage voice-based assistants to facilitate strong interactions and engagement in different industries, including\nboth small and medium enterprises and large enterprises.\n\n \n\n(2) *Architectural Design\nAI service*s We provide professional architectural drawing review solutions. By using computer vision, natural language processing\ntechnology and our unique map, image morphology processing, pattern recognition, image segmentation, image target detection, path planning,\nOCR and many other independent research and development technologies, combined with the rich professional experience in architectural\ndesign, we have launched AI products for blueprint review to achieve automation and intelligence, enabling the architecture industry to\nreduce the cost of reviewing blueprints, improving the efficiency, and cross-institution collaborative drawing review.\n\n \n\n(3) *Smart City* We use\nnatural language processing, data intelligence and other technologies to build a cognitive brain for smart city public services, and continuously\nimproves the level of urban intelligence from social service efficiency and public experience. We provide solutions such as smart city\nservice hotline, smart public service and smart legal services.\n\n \n\n**With the introduction of Hua Zang LLM in\nJune 2023, we reformed our product line.**\n\n \n\n**Hua Zang LLM: Ecosystem and LLM Commercialization**\n\n \n\n**Hua Zang LLM**\n\n \n\nThe LLM is not a novel entity; rather, it represents\nan integrated system that combines a variety of cutting-edge technologies with established AI technologies. Throughout the evolution of\nLLMs, numerous traditional AI techniques have been consolidated and refined to support the construction and application of these models,\nincluding but not limited to machine learning, deep learning, knowledge representation and reasoning, natural language processing, pattern\nrecognition, reinforcement learning, and transfer learning. With over two decades of technology accumulations in the field of cognitive\nintelligence, Xiao-I’s transition from traditional AI products to LLM product a natural progression.\n\n \n\nIn the year 2023, our Company has reaffirmed our\nposition at the forefront of technological innovation with the introduction of the “Hua Zang Universal Large Language Model”\non June 29, 2023. This model is a comprehensive AI solution that encompasses a vast array of capabilities, setting a new standard in the\nindustry with its core attributes of being controllable, customizable, and deliverable. It addresses prevalent challenges within the global\nAI landscape with innovative solutions.\n\n \n\nLeveraging the robust framework of the Hua Zang\nUniversal Large Language Model, we further expanded our technological impact with the unveiling of the transformative Hua Zang Ecosystem\non October 26, 2023. This ecosystem is poised to have a significant and lasting impact on the industry. The Hua Zang Developer Platform\nserves as the backbone of this ecosystem, offering a suite of services including development support, market outreach, and financial investment,\nthereby fostering a dynamic network that connects global partners, customers, and developers.\n\n \n\n**Hua Zang Ecosystem and Commercialization**\n\n \n\nThe “1+1+3” framework of the Hua Zang\necosystem comprises a set of foundation models, a product support platform, and three service guarantees.\n\n \n\n60\n\n \n\n \n\nThe “a set of foundation models” refers\nto the Xiao-I Hua Zang LLM, which possesses the core features of “controllability, customizability, and deliverability.” It\nincorporates technical characteristics such as deep learning models, pre-training and fine-tuning, multi-level attention mechanisms, context\nmodeling and awareness, multi-task learning, and domain knowledge integration. This enables efficient and accurate natural language processing\ncapabilities, demonstrating excellent adaptability and scalability. Leveraging its robust core technology, the Xiao-I Hua Zang LLM boasts\nhundreds of generic large model capabilities, including comprehension and generation of complex texts, mathematical reasoning, among others.\nFor instance, it can understand article information and intent, extract key information based on requirements, and rapidly analyze emotional\ntones in texts. Moreover, through continuous iteration and innovation, it achieves multi-modal capabilities such as text-to-image, text-to-edit\nimage, and image-to-text conversions. Moreover, in response to customers’ diverse business needs and budgets, Xiao-I offers a range\nof models of different sizes, enabling more flexible solutions to meet the demands of different business scenarios.\n\n \n\nThe “a product support platform” refers\nto the Hua Zang Developer Platform, comprising sections for development, applications, and operations. It assists ecological partners\nin developing applications based on Hua Zang LLM, conveniently training their own LLMs with lower costs, faster access and direct productization.\nThis embodies the Hua Zang ecosystem’s principles of “faster, lower, and more effective.”\n\n \n\nThe “three service guarantees” refer\nto the resource empowerment provided by Xiao-I through cultivation, market development, and investment. Hua Zang Ecosystem offers a wealth\nof courses and training services. Additionally, it offers integrated marketing and promoting services, product co-developing and media.\nFurthermore, through the Hua Zang investment platform, it assists ecological partners in achieving business growth and expanding commercial\nvalue.\n\n \n\nThe key to the sustainable development and iteration\nof LLMs lies in commercialization, and the best way to achieve commercialization is to create a complete industry application ecosystem.\nHua Zang ecosystem not only signifies a significant step forward for Xiao-I in the commercialization of LLMs but also marks an important\nexploration in the commercialization path for the entire artificial intelligence large model industry.\n\n \n\n**Recent Product Developments**\n\n \n\nIn 2024, we marked a strategic leap into the consumer hardware sector\nwith the launch of our AI-powered smart glasses. This move represented not only a product innovation milestone but also a transformative\nstep toward global consumer engagement. The second half of 2024 saw the successful overseas debut of these smart glasses, primarily targeting\nmarkets in North America and Europe. Fueled by rising demand for wearable AI technology, this launch drove an 18-fold increase in hardware\nrevenue—reaching US$1.40 million in 2024, up from US$75,363 in 2023.\n\n \n\nBuilding on the momentum of our smart glasses, we introduced tAIkbox,\nan advanced AI-powered customer service solution that integrates proprietary hardware with our advanced software platform. Designed for\nglobal deployment, tAIkbox harnesses the capabilities of our Hua Zang large language model (LLM), agentic task automation, and edge-computing\narchitecture to deliver seamless, context-aware interactions in real time. It is purpose-built for enterprise applications where efficient,\nresponsive, and localized AI engagement is critical.\n\n \n\nFrom an accounting perspective, AI glasses and tAIkbox platform follow\ndistinct classification protocols: By way of delivery, both products are\nrecorded as Sale of hardware products, reflecting their physical product form. By product lines,\nthey are aggregated under Model as a Service (MaaS) revenue streams, capturing recurring software subscriptions, LLM licensing fees, and\nedge-computing service modules.\n\n \n\nWhile initial adoption rates and partner feedback for both products\nhave been encouraging - indicating strong revenue potential in 2025 and beyond—our expansion into consumer hardware also presents\nsignificant operational challenges. We have limited experience in large-scale manufacturing, marketing, and distribution of consumer electronics.\nScaling these operations introduces risks related to supply chain management, quality assurance, cost control, production capacity, and\nafter-sales support. Any shortcomings, such as product delays, quality issues, or failure to meet consumer expectations, could adversely\nimpact our brand, customer trust, and long-term market positioning.\n\n \n\n61\n\n \n\n \n\nAccordingly, continued investment in operational\ninfrastructure, strategic manufacturing partnerships, and robust customer service systems will be essential to the success of our consumer\nhardware strategy. Failure to effectively execute in these areas could materially and adversely affect our growth trajectory, financial\ncondition, and overall results of operations. \n\n \n\n**Business Model**\n\n \n\nAs of the date of this annual report, we offer\ntwo different product lines: (i) Model as a service (“MaaS”) and (ii) non-MaaS. The MaaS product line includes the development\nand training, optimization and integration, service packaging and API design of the model, local deployments and subscription of our model\nproducts. Additionally, the MaaS product line also includes service and others, as customization is required to catering the demand of\nclients. The non-MaaS product line includes the needs assessment, solution design and architecture planning, development and configuration,\ndeployment and implementation of our non-model products.\n\n \n\nThe following table sets forth the product lines\nof our net revenues by amounts and percentages of our total net revenues for the periods presented:\n\n \n\n  \nFor the Years Ended December 31, \n\n  \n2023  \n2024  \n2025 \n\n  \nUSD  \n%  \nUSD  \n%  \nUSD  \n% \n\nMaaS \n 19,193,177  \n 32.4% \n 51,324,373  \n 73.0% \n 7,843,234  \n 63.6%\n\nNon-MaaS \n 39,972,082  \n 67.6% \n 18,989,942  \n 27.0% \n 4,484,450  \n 36.4%\n\nTotal \n 59,165,259  \n 100.0% \n 70,314,315  \n 100.0% \n 12,327,684  \n 100.0%\n\n \n\nBy way of delivery, MaaS and non-MaaS product\ncan be delivered in different ways: software delivery, embedded in hardware products, technology development services, maintenance services\nand cloud platform subscription.\n\n \n\nThe following table sets forth the components\nof our net revenues by amounts and percentages of our total net revenues for the periods presented:\n\n \n\n  \nFor the Years Ended December 31, \n\n  \n2023  \n2024  \n2025 \n\n  \nUSD  \n%  \nUSD  \n%  \nUSD  \n% \n\nSale of cloud platform products \n 47,007,556  \n 79.5% \n 40,877,256  \n 58.1% \n 4,329,729  \n 35.1%\n\nTechnology development service \n 7,839,700  \n 13.3% \n 24,105,644  \n 34.3% \n 2,147,523  \n 17.4%\n\nSale of software products \n 1,566,455  \n 2.6% \n 1,516,169  \n 2.2% \n 472,559  \n 3.8%\n\nM&S service \n 2,676,185  \n 4.5% \n 2,419,901  \n 3.4% \n 4,809,979  \n 39.0%\n\nSale of hardware products \n 75,363  \n 0.1% \n 1,395,345  \n 2.0% \n 567,894  \n 4.6%\n\nTotal \n 59,165,259  \n 100.0% \n 70,314,315  \n 100.0% \n 12,327,684  \n 100.0%\n\n \n\nOur Competitive Advantages\n\n \n\nWe believe we have the following competitive advantages\nand they distinguish us from our competitors:\n\n \n\nOur Pioneer Position in AI Technology and Focus\non R&D\n\n \n\n●We\nbelieve that we pioneered the industry’s first cognitive intelligence and narrow artificial intelligence technology and have built\non our culture of innovation.\n\n \n\n●Since\nits establishment in 2001, Xiao-I has focused on developing cognitive intelligence technologies based on its natural language processing\nand “AI” implementation in businesses, enjoying a privileged reputation in the “AI” industry. As a leading AI\ntechnology and industrialization service platform in China, through years of operation, Xiao-I has established cooperation with many\nleading companies amongst various industry verticals according to Frost & Sullivan. Our industry leadership is built on our pioneering\nresearch to commercialize AI technology.\n\n \n\n62\n\n \n\n \n\n●Our\nfirst-mover advantage in natural language processing has made us a pioneer in formulating AI industry standards and creating more than\n500 patents granted or pending. To protect its technology, in June 2012, Shanghai Xiao-I sued Apple Computer Trading (Shanghai) Co.,\nLtd., a subsidiary of Apple, Inc. for patent infringement and received the Supreme People’s Court Supreme Court Administrative\nJudgment, a final judgement confirming the validity of our patent in June 2020, but did not make a ruling on whether Apple infringed\nour patent. Specifically, according to the Patent Administration (Patent) Retrial Administrative Judgment issued by the Supreme People’s\nCourt of China ((2017) ZGFXZ No. 34), in the retrial case of Shanghai Xiao-I and Apple Computer Trading (Shanghai) Co., Ltd. and the\nChina National Intellectual Property Administration, the Supreme People’s Court determined that the invention patent named “A\nChatbot System (Patent No.: 200410053749.9)” held by Shanghai Xiao-I is a valid patent. On August 3, 2020, after obtaining the\nfinal judgment confirming the validity of its patent, Shanghai Xiao-I filed another infringement lawsuit against Apple Computer Trading\n(Shanghai) Co., Ltd., Apple, Inc., and Apple Computer Trading (Shanghai) Co., Ltd. (together, “Apple”), demanding Apple to\nstop the infringement and compensate for the losses. On August 27, 2020, the High People’s Court of Shanghai formally accepted\nthe Patent Infringement Case filed by Shanghai Xiao-I against Apple. On September 4, 2021, Shanghai Xiao-I filed a behavior preservation\napplication (injunction) with the Shanghai High People’s Court, demanding Apple to immediately stop the patent infringement involving\nSiri, including but not limited to stopping the production, selling, offering to sell, importing or using of iPhone products that infringe\nShanghai Xiao-I’s patent. On February 3, 2023, Apple filed a lawsuit against Shanghai Xiao-I with the Shanghai High People’s\nCourt, requesting (i) confirmation that the iPhone SE, iPhone 12, and iPhone 13 series products equipped with Siri (the “products\nin question”) do not infringe on the patent rights of ZL200410053749.9 invention patent, and (ii) Shanghai Xiao-I to compensate\nthe plaintiff for reasonable expenses, including lawyer fees, notarization fees, etc., totaling RMB 2 million currently. On January 29,\n2024, the Shanghai High People’s Court decided to merge the above two cases for trial (the\n“Patent Infringement Case Trial”). On July 31, 2024, the Shanghai High People’s Court officially concluded the Patent\nInfringement Case Trial. On September 24 and November 1, 2024, the Shanghai High People’s Court held two hearings on the case,\nboth Shanghai Xiao-i’s and Apple’s legal teams were in attendance for the hearing, which marks the latest step in this dispute.\nShanghai Xiao-i is now awaiting the verdict On March 27, 2023, the Beijing Intellectual\nProperty Court notified Shanghai Xiao-I that Apple Computer Trading (Shanghai) Co., Ltd. had filed a patent administrative lawsuit against\nthe defendant China National Intellectual Property Administration and the third person, Shanghai Xiao-I, regarding the 58271 and 58272\nReview Decision of Request for Invalidation. On June 28, 2024, the Beijing Intellectual Property Court ruled against Apple, upholding\nthe validity of Shanghai Xiao-i’s patent. This ruling validated Shanghai Xiao-i’s intellectual property and paved the way\nfor the subsequent trial.\n\n \n\n●We\nare a pioneer in AI +, with over 20 years of development and innovation, supported by four R&D centers, and collaboration with more\nthan 10 universities.\n\n \n\nAdvantages of Our Products and Services\n\n \n\n●We\ndevelop and commercialize Metaverse-related offerings, including Virtual Humans and AR/VR.\n\n \n\n●We\nhelp our clients with their digital transformation using our cognitive intelligence and AI technologies.\n\n \n\n●We\nenable our customers to reap economies of scale by providing one-stop shop service from our extensive network of service hubs in their\nvicinity.\n\n \n\n●Our\ndeep-rooted attention to quality assurance in our product and service offerings puts us ahead of our competitors.\n\n \n\n●We\nhave a proven monetization model based on product differentiation, revenue source diversification, and customer loyalty.\n\n \n\n●Our\nproducts and services meet the needs of different customers and we maintain frequent client engagement for continuous business development\nand customer loyalty cultivation.\n\n \n\n●While\nour customer contracts vary, they generally represent multi-year engagements, giving us visibility into future revenue. We have master\nsimilar commercial arrangements in place with many of our customers, retaining customers over the long term.\n\n \n\n63\n\n \n\n \n\n●Hua\nZang LLM’s core strength:\n\n \n\n†Faster\naccess: Quickly integrate our client’s system into the Hua Zang Ecosystem.\n\n \n\n†Lower\ncost: Conveniently access to the Hua Zang Ecosystem at a lower cost.\n\n \n\n†More\nefficient operations: AI gadgets, handy tools, and comprehensive guidance of operation provided on the platform leads to direct productization.\n\n \n\nOur Robust Ecosystem of Partnerships\n\n \n\n●We\nhave various regional sales teams including Shanghai, Beijing and Hong Kong.\n\n \n\n●We\nmaintain good relationships with suppliers that have a good record of performance.\n\n \n\n●Our\nproducts cover large and medium-sized contact centers, financial institutions, communication operators, government services, industrial\nmanufacturing, healthcare, and other customer groups.\n\n \n\n●We\nbuild strong and long-standing customer relationships with large enterprises in China. Our client list includes nearly all the industry\ngiants in the banking and telecom industries in China.\n\n \n\nOur Visionary and Seasoned Management\n\n \n\n●Our\nCEO Mr. Hui Yuan is a recognized AI industry Key Opinion Leader and domain expert.\n\n \n\n●Our\nteam has deep technical expertise and proven track record of constant innovation.\n\n \n\n●We\nhave proven ability to attract and retain highly qualified talent.\n\n \n\nChallenges and Opportunities\n\n \n\nUnique challenges and opportunities are presented\nfor us to achieve continued growth in sales in each of the customer industries in which we operate.\n\n \n\n*Challenges, generally we found:*\n\n \n\n●In\nthe contact center industry, high labor costs and the requirement for continuous improvement create constant margin challenges. The low\ngross profit can also cause a decline in service quality, which limits the innovation ability of the industry.\n\n \n\n●In\nthe financial industry, banks lack AI technology capabilities and independent wholly owned technology subsidiaries.\n\n \n\n●In\nthe architecture industry, the degree of digitization is low, the architectural knowledge system is unstructured and the digital drawing\nreview is a mere formality. A lot of manpower, material and financial resources are wasted.\n\n \n\n●In\nMetaverse, there are great differences in technical paths and the product form and the industry is far from mature. Many concepts have\nnot reached an industry consensus.\n\n \n\n●In\nthe manufacturing industry, many companies lack information technology talent and coordination and integration ability across departments,\nfields and enterprises.\n\n \n\n64\n\n \n\n \n\n●In\nthe healthcare industry, the level of information technology talent in urban and rural areas is unevenly distributed. Between urban and\nrural areas, the number of health technicians in cities is almost twice that in rural areas.\n\n \n\n●In\ncity public services, the traditional urban public service supply model cannot meet new requirements of modern residents for the convenience,\nspeed, efficiency and real-time urban government public services.\n\n \n\n●Technical\nChallenges, Talents and Investment: The development of large models necessitates a high level of specialized knowledge and technology,\nnecessitating significant resource investments by companies to attract and cultivate top-tier talent. The research and development of\nlarge models involve substantial capital expenditures for the purchase and maintenance of high-performance computing equipment, storage\nsystems, and other related infrastructure. As the scale of the models increases, the costs of training and deployment also continue to\nrise.\n\n \n\n●Compliance\nIssues: With the widespread application of large models, governments worldwide are increasingly strengthening their regulation of AI\ntechnology. Companies need to comply with evolving laws and regulations to avoid potential compliance risks. Large models may raise ethical\nissues such as privacy breaches, data security, and intellectual property rights. Companies must establish rigorous data management and\nprotection mechanisms to maintain user trust and corporate reputation.\n\n \n\n●Security\nConcerns: Large models may be vulnerable to various security threats, including but not limited to adversarial sample attacks, backdoor\nattacks, and membership inference attacks. This requires companies to invest additional resources in enhancing the security of their\nmodels. As large models are applied in critical areas such as autonomous driving and financial transactions, the requirements for security\nwill further increase.\n\n \n\n●Market\nAcceptance and User Education: The market acceptance of large model technology is not yet fully determined. Companies need to invest\nresources in market education to raise user awareness and acceptance of the value of large models. Users may have overly high expectations\nfor large models, but they may encounter performance limitations and errors in practical applications. Companies need to set reasonable\nuser expectations and provide transparent communication.\n\n \n\n●As\nan AI solution company, we also face many other challenges. For example: (i) the AI industry is highly competitive, OpenAI, Microsoft,\nMeta, Baidu, Alibaba and Tencent are all in this field, (ii) urban public services cover a wide range of areas which makes it difficult\nto fully and deeply understand customers’ businesses and their needs and (iii) the company’s investment may be insufficient.\n\n \n\n*Opportunities*\n\n \n\n●These\nchallenges have created tremendous and growing market opportunities for artificial intelligence solution services in China. We believe\nwe are well-positioned to capture the growing market opportunities due to the infrastructure we have created. Our platform products and\nservices are marketed and sold primarily to customers in the following industries: (1) Contact Center, (2) Finance, (3) Urban Public\nService, (4) Construction, (5) Metaverse, (6) Manufacturing and (7) Smart Healthcare. With the development of Hua Zang Ecosystem, Hua\nZang LLM is expected to further expand the industry application in the future.\n\n \n\n65\n\n \n\n \n\nThe following diagram shows the estimated market\nsize of the artificial intelligence market in 2026 in China, according to Frost & Sullivan:\n\n \n\n \n\nOur Solutions\n\n \n\nWe provide our AI solutions and services to the\nfollowing industries:\n\n \n\nIn the contact center industry, we provide internet\nservice intelligent solutions, hotline intelligent solutions and artificial intelligence solutions. Based on the scale and characteristics\nof customers, we have launched an enterprise-level model for large enterprises and an intelligent cloud contact center service model for\nsmall and medium-sized enterprises.\n\n \n\nIn the financial industry, we provide intelligent\ncustomer service with a 24-hour uninterrupted intelligent question-and-answer function. Leveraging our natural language understanding\nand speech recognition technology, our solutions can address the real-time online question-and-answer needs of different types of financial\ninstitutions and different types of customers.\n\n \n\nFor marketing professionals, we provide intelligent\nmarketing services, continuously analyzing user data and cluster user characteristics to form user portraits.\n\n \n\n66\n\n \n\n \n\nWe also cooperate with large, medium, and small\ndomestic insurance companies to provide intelligent insurance consulting services and provide independent service functions such as business\nconsulting and business queries for insurance users.\n\n \n\nIn the architecture industry, we provide professional\narchitectural drawing review solutions. By using computer vision, natural language processing technology and Xiao-I’s unique map,\nimage morphology processing, pattern recognition, image segmentation, image target detection, path planning, OCR and many other independent\nresearch and development technologies, combined with the rich professional experience in architectural design, we have launched AI products\nfor blueprint review to achieve automation and intelligence, enabling the architecture industry to reduce the cost of reviewing blueprints,\nimproving the efficiency, and cross-institution collaborative drawing review.\n\n \n\nIn Metaverse, Xiao-I invented intelligent robots\nand virtual humans, which are widely applied in various business scenarios such as exhibitions, customer services, property management,\ncare and companionship, and transaction processing.\n\n \n\nIn the manufacturing industry, we provide intelligent\nresearch and design, intelligent production process, intelligent logistics management, intelligent marketing service and intelligent management.\n\n \n\nIn the healthcare industry, we provide intelligent\nhospital services, with all-round services for patients before, during and after diagnosis. We also provide smart clinic services with\nauxiliary decision-making and interdisciplinary diagnosis and treatment and intelligent scientific research services. Based on the patient’s\ncondition, our robots generate a model to predict the clinical events, and automatically query similar cases and diagnosis and treatment\nplans in the clinical case database for doctors’ reference, providing real-time support for the doctor’s diagnosis process.\nWe connect patients, doctors, experts and medical record managers via personal computers, tablets, mobile phones, and other collaborative\nstandard videos, breaking the distance barrier and enabling them to provide patient-centered care through video anytime and anywhere.\n\n \n\nIn city public services, we integrate urban service\nresources, providing urban residents with a multi-modal human-computer interaction interface on all media channels, and comprehensively\nimprove the intelligence level of urban services.\n\n \n\nOur Growth Strategy\n\n \n\nWe intend to achieve our mission and further grow\nour business by pursuing the following strategies:\n\n \n\n●**Continue\nto improve cognitive technology capability**. We have set up a technology research institute to conduct in-depth communication on technological\ninnovation with experts and scholars from top universities, such as Duke University, Hong Kong University of science and technology and\nColumbia University. We have also carried out in-depth cooperation with well-known domestic universities to jointly develop the latest\nand cutting-edge technologies.\n\n \n\n●**Further\ndevelop and create long-term sustainable commercialization opportunities through technology innovation, application combination innovation,\nand AI product diversification.**For example, our commercialization in the field of intelligent drawing review has met the needs of\nthe construction industry for drawing review through our artificial intelligence technology.\n\n \n\n●**Further\nstrengthen the leading position in the metaverse related products**. We began to design and produce a virtual human in 2016. Our first\nmover advantage in the metaverse will help us continue to succeed in this field.\n\n \n\n●**Expand\nour customer base and make full use of existing customers through market segmentation and personalization.** We will gradually expand\nour target customers from the previous major customers to small and medium-sized customers, to provide services for a wider range of\ncustomer groups.\n\n \n\n●**Increase\nhardware products**. As a company mainly engaged in software sales and services, we will increase integrated software and hardware\nproducts in the future.\n\n \n\n●**Further\nexpand our global footprint strategically**. The goal of the company is to become a global artificial intelligence enterprise. We are\ncommitted to internationalizing our products and services and providing high-quality products and services to customers around the world.\n\n \n\n●**Further\nexpand our product line of business to customers**. With the development of LLM, the application prospects for consumer products are\nundergoing significant changes: natural user interactions, better comprehension and reasoning capabilities and wider range of application\nscenarios. Since the launch of the Hua Zang LLM, the company has been developing and expanding its 2C business, aiming to bring the benefits\nof AI to a wider range of consumers.\n\n \n\n67\n\n \n\n \n\nOur Customers\n\n \n\nWe provide our products and services to hundreds\nof enterprises across various industries, including contact center, financial sector, government and healthcare. Our customers include\n80% of the top 10 banks in China in terms of asset size, as well as 60% of the top 10 insurance companies in China. Our customers also\ninclude many leading enterprises in aviation, automobile, logistics, computer, communication, consumer and other industries and China’s\ntop 500 enterprises.\n\n \n\nFor the year ended December 31, 2023, our\ntotal sales to our top 3 customers accounted for 29.3%, 12.4% and 10.6% of our revenues, respectively. For the year ended December 31,\n2024, our total sales to our top three (3) customers accounted for 22.4%, 9.3% and 6.7% of our revenues, respectively. For the year\nended December 31, 2025, our total sales to our top three (3) customers accounted for 31.9%, 14.8% and 11.8% of our revenues, respectively.\n\n \n\nIn connection with the License Agreement, the\nVIE and China Construction entered into an Intelligent Drawing Platform Operation and Technical Support Agreement (the “Support\nAgreement”) pursuant to which the VIE has agreed to provide technical support and co-operation of the Drawing Platform for a term\nof three years. During the term of the Support Agreement, China Construction, with the consent of the VIE, may license the use of the\nDrawing Platform to third parties. In such event, the VIE shall be entitled to receive 30% of the license fee paid to China Construction\nby any third party. The technical support service was completed in 2022. The revenue recognized of technical services for the Drawing\nPlatform was nil in 2023, 2024 and 2025, respectively.  \n\n \n\nOur Suppliers\n\n \n\nWe maintain good relationships with suppliers\nthat have a good record of performance. Beijing Blanstar Technology Co., Ltd., a company established and existing under the laws of the\nPRC (“Blanstar”), was the VIE’s major service provider for the year ended December 31, 2023, 2024 and 2025 accounting\nfor 37.6%, 24.4% and 21.5% of the Company’s total purchases, respectively.  Pursuant to the terms of the Cloud Computing Technical\nServices Cooperation Agreement effective as of January 1, 2021 (the “Services Agreement”) between the VIE and Blanstar, Blanstar\nagreed to provide the VIE with cloud computing technical services consisting of various products and services including computing, storage,\nnetwork, security, management and cloud database to meet the different needs of the VIE’s various websites, applications and other\nproducts and services. Blanstar agreed to provide the VIE with response, technical support and maintenance services 24 hours a day 7 days\na week. The Service Agreement renewed on April 19, 2024 and expired December 31, 2025, subject to the right of the parties to negotiate\na renewal one month prior to the expiration date.\n\n \n\nFor the year ended December 31, 2023, we\nhave three significant suppliers including Blanstar, representing 37.6%, 20.7% and 14.9% of our total purchase, respectively. For the\nyear ended December 31, 2024, we have three significant suppliers including Blanstar, representing 24.4%, 9.7% and 5.5% of our total purchase,\nrespectively. For the year ended December 31, 2025, we have three significant suppliers including Blanstar, representing 21.5%, 18.5%\nand 16.1% of our total purchase, respectively.\n\n \n\nMarketing and Sales\n\n \n\nWe have built our Xiao-I (Chinese: 小i机器人)\nbrand through multitude sales channels, including:\n\n \n\n●industry\ntrade shows,\n\n \n\n●academic\nseminars,\n\n \n\n●publicity\nof major milestones and achievements, and\n\n \n\n●collaboration\nwith relevant academic, governmental and industrial parties.\n\n \n\n68\n\n \n\n \n\nWith these approaches, we have successfully built\nour brand and expanded customer markets. Our software business has experienced steady growth during the past few years.\n\n \n\nDistribution Network\n\n \n\nWe sell our products and services to end customers\nthrough our sales ecosystem, which consists of multiple regional sales teams and maintain strong relationships with suppliers. Our products\ncover large and medium-sized contact centers, financial institutions, communication operators, government services, industrial manufacturing,\nmedical care and other customer groups. Since 2015, we have launched the partner development plan, with more than 600 customers at present.\nPartners include finance, government, healthcare, energy, education, and manufacturing customers from Beijing, Shanghai, Guangzhou and\nShenzhen. Partners are divided into three levels: strategic level, commercial level and ecological level. Cooperation above the commercial\nlevel has the application scenarios of integrating the solutions of both parties and customers. We provide partners with a series of end-to-end\nsupport and services from sales support, event promotion, media publicity, training and certification and follow-up guarantee.\n\n \n\nIntellectual Property\n\n \n\nWe establish and protect our intellectual property\nrights through patent, copyright, trademark and trade secret laws, as well as non-competition, confidentiality and other contractual clauses,\nto establish and protect our intellectual property rights.\n\n \n\nAs of May 1, 2026, we have applied for 662 patents,\n368 of which have been granted and have obtained, 246 registered trademarks and 146 computer software copyrights. Also see “Item\n5.A Operating and Financial Review and Prospects – Operating Results – Major Factors Affecting Our Results of Operations\n– Intellectual Property.” We have led and participated in the formulation of 1 international standard, 5 national standards\nand 2 association standards, led the world’s first international standard for AI emotional computing, and published more than 20\nclass A papers (Class A papers refer to papers in authoritative core journals, indexed by the internationally accepted SCIE, EI, ISTP,\nSSCI and A&HCI retrieval system) every year. In June 2020, the company passed the national intellectual property management system\ncertification and obtained the certificate. This certificate represents that the company’s intellectual property management system\nconforms to the GB/T 29490-2013 standard. We applied in 2004, and in 2009 were authorized a patented technology numbered ZL200410053749.9\n(a chatbot system), which represented the world’s leading level of intelligent voice at that time.\n\n \n\nIn addition to the foregoing protections, we generally\ncontrol access to and use of our proprietary and other confidential information through the use of internal and external controls. For\nexample, for external controls, we enter into confidentiality agreements or agree to confidentiality clauses with our advertising customers\nand mobile device manufacturers, and, for internal controls, we adopt and maintain relevant policies governing the operation and maintenance\nof our IT systems and the management of user-generated data.\n\n \n\nOur Research and Development\n\n \n\nWe maintain that robust R&D capabilities form\nthe cornerstone of our innovation engine, particularly as generative AI accelerates technological evolution at unprecedented speeds. In\n2024, with foundational AI models achieving human-parity in code generation, we’ve strategically realigned our workforce to harness\nhuman-AI synergy. By implementing AI Co-pilots across development workflows, we’ve transitioned from manual coding to AI-enhanced\narchitectural design, enabling efficiency gain in solution iteration.\n\n \n\nAs of May 1, 2026, our restructured team consisted\nof 58 employees, including 27 research and development specialists, representing 46.55% of our workforce, with 11.11% of them holding\nadvanced degrees. s. A large number of our senior engineers have more than 10 years’ experience in the computer, Internet and AI\nindustries, and we also use part-time experts from several universities and research institutes. We have established joint laboratories\nwith the Institute of Software of the Chinese Academy of Sciences, East China Normal University, Hong Kong University of Science and\nTechnology, and have established in-depth cooperative relations with Tsinghua University, Fudan University, Shanghai Jiaotong University,\nBeijing University of Posts and Telecommunications and Peking University. \n\n \n\n69\n\n \n\n \n\nCompetition\n\n \n\nThe competition in the AI services industry is\nintense. We compete with various integrated AI services providers in chatbots and personal assistants as conversational intermediates.\nWe also compete with new companies entering into the AI service industry. The rapid nature of new technologies emerging also enhances\nthe competitive nature of our industry. Among the many other Chinese competitors, our products’ global competitors include Apple\nSiri, Microsoft Cortana and Amazon Echo. To gain market share, we have built good customer relationships with major banks and government\ndepartments in China. In addition, we also seek customers from different industries to maintain a long-term collaboration relationship.\nAlso see “Item 5.A Operating and Financial Review and Prospects – Operating Results – Major Factors Affecting Our Results\nof Operations - Competition.”\n\n \n\nC. Organizational Structure.\n\n \n\nThe following diagram illustrates the corporate\nlegal structure of Xiao-I as of the date of this annual report.\n\n \n\n \n\n70\n\n \n\n \n\nThe following diagram illustrates the ownership\nof the VIE, Shanghai Xiao-I as of the date of this annual report.\n\n \n\n \n\n**Compliance with Foreign Investment**\n\n \n\nAll limited liability companies formed and operating\nin the PRC are governed by the Company Law, which was amended and promulgated by the SCNPC on December 29, 2023 and came into effect on\nJuly 1, 2024. Foreign invested enterprises must also comply with the Company Law, with exceptions as specified in the relevant foreign\ninvestment laws. Under our corporate structure as of the date of this annual report, 100% of the equity interests of Zhizhen Technology\nare entirely and directly held by our Company through Xiao-I Technology Limited. Therefore, Zhizhen Technology, the WFOE of Xiao-I Technology\nLimited, should be regarded as a foreign-invested enterprise and comply with both the Company Law and other applicable foreign investment\nlaws.\n\n \n\n71\n\n \n\n \n\n**D. Property, Plants and Equipment.**\n\n \n\nOur current principal executive offices are located\nin Room 501, No. 363, Lane 1555, Jinshajiang West Road, Jiading District, Shanghai China. We lease offices in other cities where we operate\nwith an aggregate area of approximately 476.29 square meters as of December 31, 2025. These following facilities currently accommodate\nour management headquarters, as well as most of our sales and marketing, research and development, and general and administrative activities:\n\n \n\nLocation \nArea\n(Square Meter)  \nTerm \nUse\n\nRooms 404-405, Building 1, Tianqi Road, Jiangqiao Town, Jiading District, Shanghai \n 323.29  \nJune 18, 2025 to\nJune 17, 2027 \nOffice\n\nRoom 501, No. 363, Lane 1555, Jinshajiang West Road, Jiading District, Shanghai \n 153  \nJune 18, 2025 to\nJune 17, 2026 \nOffice"}