As the ChatGPT craze continues, Minds Lab has announced a new start with 'Mind AI'.

On the 5th, Mind AI held an opening ceremony to commemorate its name change at the aT Center in Yangjae-dong and through online channels. Mind AI introduced the specialized services offered on its Mind AI platform and unveiled its AI human, "M3."

Mind AI, which changed its mission from Minds Lab, announced its goal of “evolving into a comprehensive AI service company that provides customized AI learning (Fietune Train), AI model inference (Inference) service, and AI Human as its main businesses, and will collaborate with various AI business partners centered on the AI platform.”

'Personalized AI Learning (Fietune Train),' introduced as the main business of Mind AI, creates a customized engine by adding Mind AI's AI engine, existing datasets, and customer data to learn about GPT, voice, and face.

This is possible thanks to Mind AI's massive language model. Mind AI leverages this massive language model and provides it to customers in the form of an API that they can directly utilize. Mind AI's engine goes through three stages: first, data learning; second, data comparison and reward model learning; and finally, reinforcement and algorithm optimization.

At the event, Mind AI unveiled and demonstrated its customized Mind GPT, trained on 12 domains. Mind GPT is utilized for specialized services such as travel experts, real estate experts, and counselors. A Mind AI service representative stated, "We focus on B2B rather than B2C, and can provide services that enable optimized AI solutions for corporate clients."

The "AI Model Inference Service" supports the creation of customized AI services and their practical use through the Mind Orchestra platform. In essence, Mind Orchestra is a pipeline tool for conveniently utilizing Mind AI's engines and models. Customers can create AI pipelines by connecting AI models and APIs. Language models, such as Mind GPT, and AI models such as voice and face models are inferred individually or in conjunction in real time. Mind AI optimizes performance and speed.
A Mind AI representative demonstrated the process of actually constructing the Mind Orchestra, explaining, “Mind Orchestra is a low-code system that helps customers easily connect AI engines and enjoy one-stop service in actual operations.”

He also said, “Mind Orchestra guarantees the stability of the operating system and manages demanding GPU server management, such as AI engines, in the cloud,” and “It is optimized for end-to-end connection from front-end apps to business apps based on an infrastructure that integrates on-premise and cloud environments.”

Finally, "Human AI" provides a conversational customer interface using core engines for vision, hearing, voice, face, and natural language. This allows businesses to leverage human AI video and audio generated for each domain.

Mind AI's M1 model focused on integrating AI capabilities, while M2 added multilingual capabilities. M3, on the other hand, expanded its language capabilities by adding ChatGPT. The LLM (Language Model with Label Smoothing) model was applied to construct scenarios, and classification, summarization, question-answering, sentence generation, and chatbot functionality were added.

The Mind AI representative demonstrated M3, a human AI, providing professional consultations and making online and phone reservations. Connected via TTS and STF, the human AI can converse in a human-friendly manner and even conveys emotion through its voice. Human AI is being used as AI expert counselors, guides, instructors, announcers, stenographers, trainers, and more.

In the future, Mind AI plans to upgrade its current app development process, linking with Mind Orchestra, to enable immediate customization using app templates. Furthermore, the company announced plans to create an AI- and robot-powered office as a test bed at its Pangyo office in late June.