Integrit Unveils ‘Stella’: Pioneering AI Hardware Generating Dynamic Interactions with LLMs

SEOUL, South Korea, Feb. 20, 2024 /PRNewswire/ — AI Robotics platform specialist Integrit (Seoul/integrit.ai) announced on the 1st that it is commercializing a groundbreaking generative AI hardware device called ‘Stella’ (pictured). Stella automatically generates expressive behaviors for robots by applying large language models (LLMs) and generative AI technology in various customer service and business settings, enabling customized conversational search and interactions in multiple languages.

STELLA at Seoul Korea
STELLA at Seoul Korea

The innovative AI hardware ‘Stella’ integrates OpenAI’s ‘GPT-4’ and Meta’s ‘LLaMa’ among other LLMs, along with on-device AI vision technology. It provides optimal responses to customer queries and requests in high-noise environments and dynamically interacts with customers using various movements, enhancing the quality of responses. Stella’s unique generative movement technology, combined with embedded hyper vision and multimodal capabilities, allows it to recognize environments and customers, interact with them, and generate appropriate expressions and movements without the need for separate training.

Stella is a dedicated hardware device for AI conversation search using LLMs. It serves as a ‘business account’ and ‘information controller,’ enabling customers to interact with AI via voice search in department stores, shopping malls, hotel check-in counters, hospitals, unmanned stores, or bank desks. It connects to enterprise AI systems to provide customers with in-depth expert information at a low cost.

Stella is equipped with Qualcomm Technologies’ AP and is powered by the on-device AI robotics platform ‘AirPath®’ consisting of 5G, multi-camera, stereo depth camera, and AI vision solutions. It integrates a control platform that enhances generative inference capabilities in a single hardware and software environment, enabling it to generate language models and motions without the need for training.

Stella, the first generative AI hardware to be experimentally unveiled at a major department store in Korea, introduces a novel method to enhance the accuracy of conventional conversational AI search. It achieves this by analyzing information captured by the camera and generating prompts without the need for network assistance. The 11-inch OLED screen, equipped with noise pre-processing systems and speakers, detects user movements and interacts with a 360-degree rotatable responsive moving joint, providing personalized responses by identifying the user’s location, environment, age, and gender through embedded vision AI solutions.

The innovative AI hardware Stella provides a new form of interface that communicates by moving according to the display provided by Virtual Human. It differentiates itself from touch screen-based control interfaces with voice conversation, facial recognition, and gesture recognition.

Integrit CEO Brian CS stated, “The emergence of generative AI platforms such as GPT, LLaMa, and DALL-E has brought significant changes to the AI market, increasing productivity and efficiency in personal services and business environments. However, until now, there has been a lack of devices or methods to connect AI in public places, service spaces, and living environments where actual companies interact with customers.” He added, “Stella is expected to be an alternative, providing consumers with a high level of business intelligence experience without the high cost of AI environments or digital technology.”

Integrit will officially unveil the generative AI hardware Stella at MWC 2024 in Barcelona and plans to expand its global partnerships, including adding conversational AI search engines such as GPTs and Google’s ‘Gemini’, in line with the open market policies of LLM platforms in the future.”

View original content to download multimedia: Read More