Terminus Group Unveils AGI Agent at Web Summit Lisbon

Recently, the Web Summit regarded as "the Olympics of Tech” — concluded in Lisbon, Portugal. As part of the Science and Technology Week of Portugal, the summit attracted over 100,000 elites from the global science and technology community, including Brad Smith, President of Microsoft, Cristiano Amon, President and CEO of Qualcomm, Sarah Franklin, CEO of Lattice, a global semiconductor leader, and Kuo Zhang, President of Alibaba.com, who shared their insights on the future of technological development. Representing China's AI industry, Victor AI, Founder and CEO of Terminus Group, was invited to discuss emerging technological trends and new pathways driven by Artificial Intelligence as a Service (IaaS); besides, in collaboration with Buttons, one international luxury brand, Terminus introduced “Hali” — a multi-modality and multi-intelligence collaborative AGI Agent powered by Terminus's advanced AI capabilities.

From Generative AI to Multi-Intelligence Collaboration: AGI Agent Leads Interaction Transformation

Victor highlighted the emerging trends that originate from the fusion of AI and product aesthetics. The evolution of human-computer interaction is making a revolutionary leap from the digital age to the era of intelligence, progressing from CLI (Command Line Interface) to GUI (Graphic User Interface), and now to CUI (Conversational User Interface) and BMI (Brain Machine Interface).

The ongoing evolution in the age of intelligence has given rise to a new business model -- “IaaS (Intelligence as a Service), which has led to a series of innovative and IaaS-driven intelligent services. Among them, AI Agent stands out as a key representative. Victor believes that IaaS will move from generative AI to single AI Agent, eventually leading to a synergistic collaboration of multiple AI Agents in the future, while each sector will welcome their vertical AI Agents respectively. Such a shift urgently demands for a more advanced AGI Agent that is capable of coordinating multiple AI Agents to address the diverse demands of humanity in collaboration.

In response to this, Hali -- the first AGI Agent developed with Terminus's AI technology in partnership with Buttons—made its debut at the Web Summit, one of the world's premier technology gatherings.

AI Super-Species from Solaris

Hali takes its name from the super-intelligent character in the Polish science fiction Solaris. Unlike large language model (LLM), Hali represents a form of AGI Agent characterized by four core attributes: “human-like” cognition, “extended” memory, “perception” of the physical environment, and “collaboration” among multiple AI Agents.

1. “Human-like” Cognition: Unlike GPT models that prioritize rapid question-answering, Hali engages in human-like thinking by understanding the true intent behind user requests, exploring multiple solutions, and selecting the optimal one using various tools (e.g., web search, programming).

2. Extended and long Memory: Hali possesses enhanced memory capabilities. Hali is able to refine solutions over time for complex problems. Through an ongoing learning process of a user's preferences and thought patterns, the Terminus large modality model builds a personalized knowledge graph for the user, creating a unique knowledge base for every individual.

High-Dimensional Perception of the Physical World: Depending on the end devices deployed, Hali can perceive real-time changes in the physical world through sensor collaboration. For instance, Hali can detect climate fluctuations and suggest appropriate clothing and even accessories tailored to user preferences.

3. Multi-modality Scheduling and Multiple AI Agents Collaboration: Leveraging the cognitive and planning capabilities of its multimodal model, Hali can coordinate various types of AI Agents. Through straightforward language interaction, users can access a comprehensive system of synergistic AI agents to complete navigation, translation, event management, business analysis, and more tasks.

4. Analyses show that Hali’s semantic understanding accuracy exceeds 97%, with text chat response latency ranging from 500 to 1500 milliseconds, thanks to its extensive knowledge bases. In translation mode, Hali achieves a bidirectional translation accuracy of 96% between English and Chinese. Besides, in order to foster a more personalized interaction, Hali is trained with 36 psychological books to enhance its ability to comprehend human intentions and emotions.

Victor links the consumer products integrating Hali with the concept of “aesthetics”, a principle that aligns with Terminus's core philosophy. In his view, Terminus aspires to create “macro aesthetics” by employing AIoT technology to create a greener, smarter, and more vibrant environment for humanity.

At the consumer level, Terminus's products embody “micro aesthetics” by seamlessly connecting individuals to the physical world through user-centric designs. The essence of aesthetics extends beyond nice sensory experience; it is reflected in exceptional product performance, technological breakthroughs, and the seamless user experience that create a deeper connection with the world. This echoes with Terminus's brand commitment -- using AI to create a better life.