This blog post was originally published at Qualcomm’s website. It is reprinted here with the permission of Qualcomm
A day with Qualcomm revealed the innovations empowering new and exciting AI experiences running within devices
In the rapidly evolving world of artificial intelligence (AI), staying ahead of the curve is crucial. As a leader in on-device AI, Qualcomm Technologies, Inc. hosted a workshop that served as a meeting spot for ideas, innovations and discussions on the state and future of AI technologies. We welcomed members of the media, analysts, influencers and members of the Snapdragon Insiders community to be part of this jam-packed day, where our amazing partners also showcased some of their latest and greatest AI solutions powered by our on-device AI technologies. Here’s a glimpse into the day’s events and the groundbreaking insights shared by experts.
Morning presentations: Deep dives into AI innovation
We kicked off the workshop with an insightful keynote from Durga Malladi, senior vice president and general manager for technology planning and edge solutions (Qualcomm Technologies), which overviewed the current AI landscape, highlighting the transformative potential of the generative AI revolution and the scaling challenges it faces — including rapid growth of AI to support billions of users, increased electricity consumption and responsible AI development.
With Qualcomm Technologies as the partner of choice for leading on-device AI, shifting towards a hybrid model taking advantage of both cloud and edge device AI capabilities is critical to address these challenges, especially as on-device AI becomes both powerful and energy efficient.
After a quick break, we regrouped to dive deeper into three key areas where Qualcomm Technologies is hard at work:
- AI research: Jilei Hou, vice president of engineering (Qualcomm Technologies), outlined the company’s decade-long leadership in AI research and development, emphasizing its role in accelerating AI commercialization and scalability. He highlighted key advancements in on-device AI, such as enhanced architectures and how heterogeneous computing is enabling generative AI at the edge. Hou described how generative AI agents and orchestrators can significantly improve application capabilities and enable embodied AI. He also detailed Qualcomm’s innovations like quantization, compression and speculative decoding, which optimize AI model efficiency on edge devices. He also talked about the use of advanced techniques like low-rank adaptation (LoRA) for customized AI user experiences, underscoring Qualcomm Technologies’ commitment to advancing AI technology and performance across various platforms.
- AI hardware innovations: Vinesh Sukumar, senior director of product management (Qualcomm Technologies), summarized Qualcomm’s AI hardware innovations, particularly highlighting advancements around the Qualcomm AI Engine — how it cuts across the entire Qualcomm AI portfolio, powering platforms including mobile, PCs, automotive, internet of things (IoT) and extended reality (XR) products — and provided a deep dive into the powerful Qualcomm Hexagon neural processing unit (NPU) and the AI accelerators within it that drive all the on-device AI capabilities. Vinesh also talked about our unique Qualcomm Sensing Hub that powers always-on, low-power AI and adds personalization to the generative AI user experience. Finally, he also covered one of the hottest topics in the industry: the launch of a category of AI PCs and how Qualcomm Technologies Snapdragon X Series processors are the only solution powering Microsoft Copilot+ features today. Beyond showcasing the technical prowess of Qualcomm Technologies’ hardware, Vinesh also presented various AI use cases that our OEM partners are implementing into commercial devices built for consumers and organizations.
- AI software and tools: Siddhika Nevrekar, senior director of product management (Qualcomm Technologies), focused on the developer journey and discussed Qualcomm Technologies’ leading AI software from our fundamental Qualcomm AI Stack that serves all Qualcomm and Snapdragon platforms, to the Qualcomm AI Hub that provides developers with a one-stop shop for optimizing AI models — making on-device AI development as straightforward as possible. She was joined on stage by executives from Lantronix and FullTrack AI, who highlighted the significant advantages of using the Qualcomm AI Hub. They shared that it enabled them to run their applications directly on-device and dramatically reduced their development and deployment cycle from months to just a few days.
Afternoon activities: Practical applications and future visions
After lunch, we explored impactful AI applications with Angela Baker, vice president of corporate responsibility and chief sustainability officer at Qualcomm Incorporated. Angela moderated a panel discussion featuring representatives from Truepic, Cephable and McAfee.
Their discussion illuminated how AI is being leveraged to solve complex problems in diverse areas such as security, accessibility and overall, how we are enabling “AI for good” with these partners.
Finally, we stretched our legs, experiencing the technology first hand with interactive demo sessions. Attendees rotated through various stations, each dedicated to a specific domain like mobile, compute, XR, IoT and cloud. Domain experts provided insights and answered queries, offering a hands-on understanding of Qualcomm Technologies’ contributions to these fields.
We were thrilled to showcase innovations from several partners. Here are just a few highlights:
- Cephable showcased their app, which leverages neural networks for audio recognition, facial expression recognition and head movement tracking. This technology allows users to control Windows and Windows Apps using voice, face and motion, enhancing accessibility for a more inclusive digital experience. Cephable also utilized the Qualcomm AI Hub, which facilitated the seamless integration of their model directly onto the device.
- McAfee presented their Deepfake Detector, an advanced tool that uses AI detection models to identify AI-generated audio in videos. This helps users navigate their digital environments more safely and verify the authenticity of the content they encounter — all enhanced by the powerful capabilities of the Snapdragon X Elite platform.
- LM Studio introduced a version of their popular application to run large language models within PCs, now optimized to run local AI workloads with Llama 3, Phi-3, and other open-weights LLMs directly on Copilot+ PCs powered Snapdragon X Series processors.
- Lantronix showcased a demo of their Percepxion IoT edge platform, specifically designed for smart-city applications. This platform integrated Qualcomm AI Hub, enabling developers to easily optimize and deploy AI models across devices equipped with Qualcomm Technologies’ chipsets. This collaboration allows developers to focus more on creating solutions for traffic monitoring, congestion management and public safety, while Lantronix handles the complexities of the underlying AI infrastructure.
- Dynamo AI, a company specializing in AI performance, security and compliance solutions, showcased their innovative application, Dynamo Guard. It’s designed to provide robust guardrails for large language models (LLMs) and is crucial for enterprises aiming to deploy generative AI safely and effectively, ensuring that the AI operates within set ethical and operational boundaries.
- TruePic showcased how their technology empowers devices to securely verify authentic original images or to create synthetic media with complete transparency. This capability is facilitated through the use of the Coalition for Content Provenance and Authenticity (C2PA) standards, ensuring that each piece of content can be traced back to its origin.
As you can probably tell, Qualcomm’s AI Media & Analyst Workshop was a comprehensive showcase of the breadth and depth of on-device AI technologies across various sectors. It not only highlighted the company’s leadership in on-device AI but also showcased a strong ecosystem of partners and use cases that are making on-device AI a reality.
Rodrigo Caruso Neves do Amaral
Staff Marketing Communications Specialist, Qualcomm Technologies, Inc.
