AAEON Technology has collaborated with Darwin Precision Industry and JTB Tech to develop and bring to market a generative AI recognition system.
Darwin Precision Industry is a longstanding figure in optical design and precision processing technology, while JTB Tech has made a name for itself by developing advanced, customized AI solutions for multiple vertical markets, making both ideal collaborators for the project. This collaboration marks the first use of an AAEON product in a generative AI solution, representing a significant milestone for the company.
The system is currently installed in various public locations such as metro stations, train stations, and airports. The presence of generative AI products in the market is limited, especially those capable of integrating both voice and image recognition into their operation. Most existing smart customer service systems require manual language selection and depend on cloud APIs (e.g., OpenAI), leading to an industry average of a three-second delay in response time due to network latency. Moreover, the vast amount of content drawn from and generated by large language models (LLMs) can result in even longer response times. Existing solutions’ dependence on network connectivity also introduces inevitable cybersecurity risks. Companies using these systems face privacy exposure risks, and if the network fails, the system does too.
However, the generative AI recognition system developed in collaboration with Darwin Precision Industry and JTB Tech was designed to address these challenges. It not only automatically recognizes languages (including sign language) but provides responses within one second, giving users immediate feedback in a way that fully demonstrates the benefits of generative AI. The system connects to a private cloud, ensuring data security. The type of LLM employed, a “Retrieval-Augmented Generation (RAG)”, also contributes to its ability to issue faster responses. Comprehensive VPN protection ensures privacy, while the system can also operate offline, eliminating the need for a network connection.
AAEON provided a fully customized system featuring a 13th Generation Intel® Core™ i9 CPU, a water-cooled GPU, and a bespoke heat dissipation system to enhance CPU cooling. Further, the system included four 2.5” HDD bays for storage, two high-speed Ethernet ports, one USB Type-C port, and two USB 2.0 ports, as well as seven expansion slots for future scalability as LLM technology evolves.
“Most smart customer service systems on the market have limited scalability, but Darwin’s smart customer service system can be customized with IP characters, which automatically greet and attract people in public places. It can also integrate with social media, not only for advertising but also for extending business services. Furthermore, users can take selfies and download a QR code to save the files using its fun photo frame,” said Darwin Precision Industry’s Senior Director, Hsin-I Tsou. “This technology uses a large language model for fine-tuning based on user needs. It combines two multimodal translation AI systems, OpenAI and Seamless M4T, to provide real-time translation. Additionally, the system uses ASR/TTS text-to-speech conversion technology to encode and decode models. As such, each model combines the best elements of each aspect for the best possible customer experience!” Director Tsou added.
“Darwin Precision chose to collaborate with AAEON not only because AAEON’s products are stable and reliable, but due to the synergy of our collaboration with JTB Tech, which allowed us to meet their needs in a fast manner. Moreover, the most time-consuming component in product development, the graphics card, was something that AAEON anticipated the project would require, so AAEON proactively dedicated the necessary resources. As a result, this fully customized product was developed within just four weeks and delivered as a sample to Darwin. This is twice as fast as the standard eight-week product customization timeline, allowing the product to hit the market quickly,” said AAEON Sales Manager, Nina Sun.
