Agora and Seeed Studio partner to power voice-native Embodied AI with Reachy Mini at NVIDIA GTC 2026. March 18, 2026 Santa Clara, CA - March 18, 2026 Artificial intelligence is entering a new phase as it moves from cloud software into robots, edge devices, and physical environments. The global embodied AI market is projected to grow from USD 4.44 billion in 2025 to USD 23.06 billion by 2030, highlighting the rapid rise of intelligent machines capable of real-world perception, reasoning, and action. In this new generation of AI systems, voice is emerging as the primary interface between humans and machines. At NVIDIA GTC 2026, Agora, a global leader in real-time engagement and conversational AI infrastructure, and Seeed Studio, a global open hardware innovation platform providing developer tools, hardware modules, and manufacturing services for IoT, robotics, and edge AI, announced a collaboration to accelerate the development of voice-native embodied AI systems. In coordination with ecosystem partners NVIDIA, Hugging Face, and Pollen Robotics, the initiative provides developers with a complete stack for building real-time AI robots and voice-native devices, combining: * Reachy Mini humanoid robot * NVIDIA Jetson Orin Nano edge AI computing * Hugging Face multimodal open models * Agora's Conversational AI Engine By integrating robotics hardware, edge AI computing, multimodal AI models, and ultra-low-latency conversational AI infrastructure, the collaboration enables developers to bring AI agents into the physical world with ultra-low-latency interaction, global connectivity, and advanced voice capabilities such as far-field audio capture, speaker recognition, and customizable wake-word activation for natural, lifelike experiences. The collaboration will be showcased at the Seeed Studio Embodied AI Hackathon, in Santa Clara on March 21-22, immediately following GTC. Bringing digital AI to life with Reachy Mini. At the center of the initiative is Reachy Mini, an open-source humanoid robot originally developed by Pollen Robotics in collaboration with Hugging Face and integrated with Seeed Studio's hardware ecosystem. "AI is moving rapidly from cloud applications into robots and intelligent devices," said Tony Wang, Chief Revenue Officer and Co-founder at Agora. "Real-world AI requires natural, real-time interaction. Through our collaboration with Seeed Studio and ecosystem partners such as NVIDIA, developers can combine Agora's conversational AI infrastructure with platforms like Reachy Mini and edge computing to build voice-native intelligent agents designed for real-world environments." During the hackathon, developers will build embodied AI agents by combining the Reachy Mini robot platform, NVIDIA Jetson edge AI computing, ElevenLabs STT, multimodal AI models from Hugging Face, and Agora Conversational AI for real-time, human-like interaction. This architecture allows advanced multimodal AI models to run seamlessly on edge hardware while enabling natural, real-time human-robot interaction through voice. The result is a new generation of AI-native robots capable of perception, reasoning, and real-time conversation, enabling new possibilities for personal robotics, intelligent assistants, and AI companions. Expanding voice-native AI devices with reSpeaker. In addition to robotics innovation, Agora and Seeed Studio are collaborating to enable voice-native edge AI devices through Seeed Studio's reSpeaker platform. Powered by the XMOS XVF3800 AI voice processor, the reSpeaker 4-microphone array provides advanced far-field voice capture capabilities, including: * 4-microphone circular array with 360° far-field voice pickup (up to 5m) * On-board AI acoustic Algorithms (Beamforming, AGC, AEC, DoA, Noise Suppression) * Seamless integration for deployment * Plug-and-Play simplicity and compact design for embedded applications These capabilities allow voice AI systems to operate reliably even in noisy environments across various applications, such as smart homes, smart conferencing, AI voice agents, and robotics. When integrated with Agora's Conversational AI Engine, the reSpeaker platform becomes a complete edge conversational AI system. The architecture enables a real-time conversational loop where: * The reSpeaker microphone array captures user speech * Audio is transmitted through Agora's Software-Defined Real-Time Network (SDRTN(R) * Agora's Conversational AI Engine performs speech recognition, AI reasoning, and TTS * Responses are streamed back to the device and played through speakers This end-to-end pipeline enables developers to build production-ready voice AI devices such as smart assistants, robotics interfaces, AI companions, and intelligent IoT products. Leslie Liao, Vice President of Product at Seeed Studio, added, "The seamless integration of Agora's Conversational AI Engine (Convo AI) and Seeed's reSpeaker serves as the smart ear for embodied AI, enabling robots and intelligent hardware to interact with humans through natural, real-time conversation. Together with our open hardware ecosystem and partners like NVIDIA and Hugging Face, we are empowering developers to build the next generation of embodied AI applications." Advancing multi-modal physical intelligence. To highlight the shift into the next generation of physical AI, Agora will present the Convo AI Embodied Intelligence Award during the hackathon. The award recognizes teams building voice-native, multi-modal physical intelligence designed for real-world deployment. Winning projects will demonstrate systems where: * Voice serves as the primary human-robot interaction layer * Multi-modal intelligence enables real-time perception and action * The hardware abstraction layer enables cross-chipset scalability * Embodiment is engineered for industrial readiness From prototype to platform, the award celebrates innovations advancing the next generation of physical AI systems built to scale. Empowering the Embodied AI developer community. The Embodied AI Hackathon invites developers, engineers, and robotics enthusiasts to explore the future of AI-powered robotics and voice-native devices. Participants will receive hands-on access to: * Reachy Mini Lite robots * Agora Conversational AI SDK * Open-source AI models and frameworks * Technical mentorship from ecosystem partners Projects will be evaluated on technical implementation, innovation, and completeness, with prizes totaling over $6,000, featuring high-value coupons, a Reachy Mini Lite, and Amazing Hand. About Agora. Agora is a global leader in real-time engagement, providing developers with the APIs to embed real-time voice, video, and conversational AI into any application. Headquartered in Santa Clara, CA, Agora is trusted by over 2,000 organizations to power best-in-class digital experiences across all industries. About Seeed Studio. Seeed Studio has been a leading Open Hardware company since 2008, empowering users to create real-world digital solutions. Agora provide industrial-ready modules and devices, and open up the capability of product, production, and promotion. By embracing open source, community building and integrated software suites, Agora is proactively lowering the tech barriers and including users with diverse expertise for localized matters. 2804 Mission College Blvd. Santa Clara, CA, USA 95054 GET STARTED

Agora enables scalable Voice AI agents for enterprises. Search. SDRTN(R)-powered agents eliminate latency and production hurdles, making Conversational AI finally deployable at scale Agora (NASDAQ: API), a global leader in real-time engagement, today unveiled its Conversational AI Agent solutions for Customer Service and Sales & Marketing - built on a unified platform designed to solve the technical friction that has kept Conversational AI from true enterprise-grade deployment. This arrives as the industry reaches a critical inflection point. According to Gartner, by the end of 2027, conversational AI agents are expected to automate 70% of customer interactions, and by 2028, AI agents are projected to outnumber human sellers 10 to 1. Despite this massive opportunity, many businesses remain constrained by legacy systems that are simply unable to address long wait times, rigid call routing, or inefficient outreach - creating operational strain while eroding customer trust. The Infrastructure Powering Enterprise Voice AI Agora's comprehensive Conversational AI platform eliminates high-stakes production hurdles by unifying three core technologies into a single, high-performance stack: * Agent Studio: A visual, no-code environment that enables enterprises to rapidly build, test, and deploy sophisticated voice AI agents. * Conversational AI Engine: An advanced orchestration layer that integrates Automatic Speech Recognition (ASR), Large Language Models (LLMs), and Text-to-Speech (TTS) for seamless real-time workflows. * SDRTN(R)(Software-Defined Real-Time Network): Agora's foundational, globally distributed infrastructure that ensures sub-second latency and reliability. By leveraging this high-performance stack, Agora delivers a level of operational reliability that competing voice AI solutions cannot match - ensuring natural conversational flow through ultra-low-latency interactions, crystal-clear communication on congested networks, and accurate responses in noisy environments through AI noise suppression and voice locking. "Voice is the most natural way we communicate, but delivering real-time conversational AI at scale requires infrastructure built specifically for it," said Tony Zhao, Founder and CEO of Agora. "By solving the latency and orchestration challenges that have historically limited voice AI, we're enabling companies to finally deploy conversational agents at scale without compromising on the quality of the customer experience." Currently, global organizations are already deploying these capabilities to transform high-volume customer service and outbound engagement workflows. Transforming Customer Conversations with Voice AI AI Customer Service Agents: Scaling Support Without Compromising Experience As customer interaction volumes increase, contact centers face mounting pressure to maintain service quality while managing operational costs. Agora's AI Customer Service Agents solve this by automating routine inquiries - such as appointment reminders, shipping updates, technical troubleshooting, and billing inquiries - providing consistent, real-time support that allows human agents to focus on more complex cases. Key advantages: * Instant availability: Provide 24/7 automated support to resolve common inquiries without waiting. * Natural conversations: Facilitate fluid, human-like dialogue using advanced interruption-aware voice technology. * Operational efficiency: Maximize resolution rates through intelligent automation and seamless escalations to live human agents when needed. AI Sales & Marketing Agents: Intelligent Outreach at Scale Outbound engagement has historically relied on rigid dialers and scripts that feel impersonal to customers. Agora's AI Sales & Marketing Agents solve this by bringing conversational intelligence to outreach - powering high-impact use cases like debt collection, payment processing, outbound lead qualification, interactive surveys, and real-time event polls - through natural, real-time voice interactions. Key advantages: * Context-aware engagement: Navigate complex objections and adapt messaging dynamically in real time. * Automated qualification: Identify high-intent buyers and schedule qualified meetings directly. * Higher productivity: Enable sales teams to focus on closing deals rather than manual dialing and screening. "The challenge with traditional market research is the friction of the initial outreach. By using Agora's Voice AI agents to invite participants to our survey platform, we've seen a 10% conversion rate - a figure that was previously unattainable with manual dialing," said Alejandro Pinto Villalba, Executive Director at FasesBI. "The agents handle the initial pitch and incentive explanation flawlessly, allowing us to scale our data collection and reward distribution without increasing our operational headcount."

Agora reported its fifth consecutive quarter of GAAP profitability in Q4 2025, marking its first full profitable year since 2018. The real-time engagement platform generated $38.2 million in revenue, up 10.7% year-over-year, with GAAP net profit of $4.9 million and a 12.9% net margin. The company demonstrated its platform capabilities during a Super Bowl weekend live streaming event for MrBeast on OneNote, delivering full HD video to nearly 600,000 peak concurrent viewers with sub-second latency—described as the largest live video shopping event in US history. Agora's conversational AI engine, launched in March 2025, has seen usage more than double each quarter. The technology is being deployed across customer service, smart devices, education and AI-powered consumer applications, with strong adoption in companionship toys.