*Originally published in Chinese on Central News Agency on April 28, 2023
Taiwanese AI startup Aiello announced today that their smart speaker, Aiello Voice Assistant (AVA) is expected to expand its language capabilities with large language models (LLMs).
AVA becomes the world’s first smart speaker to integrate voice and the ChatGPT language model. AVA is the core product by Aiello, and it was first launched in 2020. The AI assistant was first used in the hotel industry to assist guests in solving a wide range of requests. Currently, it has been introduced in over 60 hotels and 10,000 rooms in Taiwan, Japan, Thailand, Singapore, and Malaysia.
At the recent press conference organized by Aiello, the CEO Vic Shen said that the company is committed to development in the field of natural language technologies. In 2020, Aiello team began working with OpenAI’s GPT-series of language models and were among the first five companies worldwide to be authorized by OpenAI.
Hotel guests can ask AVA their questions such as recommendations for nearby attractions or inquiring about check-out times, which helps save a significant amount of time for hotel staff. For example, after implementing the AVA hotel AI system in 200 rooms, the front desk team at InterContinental Kaohsiung was able to lower the call handling time by 20%. With AVA in the room, the hotel guests also often share their experience on social media which increases the hotel’s brand exposure.
Aiello CEO Vic Shen said that AVA will integrate large language models, including Azure OpenAI GPT-3.5 and Bloom.
This integration will make the interactions with the AI speaker more natural, assisting with tasks such as restaurant reservations and inquiries about check-out times. The large language models can provide more comprehensive responses, such as: “The check-out time is before 12 PM. If you need to extend your stay, you can ask the hotel front desk.”
This improvement is expected to increase usage rates and significantly reduce the hotel’s maintenance costs. Vic Shen explained that previously, it would take 2 to 3 months for a hotel to teach AVA AI butler to learn new information, for example about a marketing campaign. With the large language models this process is way faster. The hotel can simply upload a PDF file for the model to learn from and AVA will be able to respond to new set of questions as early as the next day.
Google Taiwan’s former managing director, Lee-Feng Chien, was also impressed by the AVA’s large language model integration. He believes that in the future, we will separate between language models designed for general use and language models designed for industry verticals.
Lee-Feng Chien stated that after integrating the large language model with AVA, the conversational content can also become valuable source of data for the company. This allows for the automation of relevant processes, which he considers a breakthrough change.
Today, Aiello officially released the Vocol Language Collaboration Platform. Apart from being able to upload audio files and convert them into transcripts, Vocol also utilizes AI to generate meeting record summaries. Currently, the platform has over a thousand registered users. Vic Shen said that Vocol can perform cross-platform analysis and online conference recordings.
The company aims to expand their user base from individual consumers to enterprises. This product is primarily targeting the customer service industry. They also aim to extend their services to industries such as telecommunications. Looking ahead to the future, Vic Shen stated that the pandemic has prompted more hotels to accelerate their transformation. Aiello aims to establish a strong presence in the Asia-Pacific region this year. The team estimates that the number of hotel rooms equipped with smart speakers will reach 20,000 this year. Furthermore, they hope to expand into the Australian market next year.
Do you want to dive deeper into conversational AI solutions for enterprises? Message our expert team and we’ll get back to you within 2 working days.
Do you want to learn more about voice AI technology? Sign up for a free demo.