"Timely Language" received tens of millions of yuan in Pre-A+ round financing, and the new version of the cross-border AI customer service product driven by the big model has been launched

Text｜Shen Xiao

Editor: Wang Yutong

36Kr learned that Beijing Timely Language Intelligent Technology Co., Ltd. (hereinafter referred to as Timely Language) has completed tens of millions of yuan in Pre-A+ round of financing, exclusively invested by Yuandu Capital. This round of financing will be mainly used for cross-border AI customer service product development and iterative optimization, and to establish a regional sales team.

Shishiyu was founded in 2021 and has previously completed angel and pre-A rounds of investment from Zero One Ventures and Jushuitan. Relying on AI technology, the company focuses on providing a multi-platform/multi-channel centralized and unified intelligent customer service SaaS platform for cross-border e-commerce sellers, aiming to meet the needs of overseas enterprises for channel acquisition, customer operation and brand management. The platform functions cover AI message automatic reply, IM chat, user emotion management, AI order management, global language translation, VOC (Voice Of Customer) data management and data analysis reports, etc.

According to Bai Tao, founder and CEO of Shishiyu, since its establishment, the company has made great progress in products, technology, and market.

In terms of products and technology, while gradually improving the platform functions, Timely has completed the integration of chat tools on the two major social media platforms, FaceBook and Instagram, as well as mailboxes such as NetEase, Yahoo, Gmail, and Outlook, and has achieved API docking with more than 10 mainstream cross-border e-commerce platforms and applications such as Amazon, eBay, Shopify, Shoplaza, SHOPLINE, and Shopyy. Customers can authorize multiple channels and stores with one click and complete centralized customer management through a unified operation interface.

On the other hand, the AI ​​functions of the company's original products are mainly developed based on the Bert model and other multimodal models. At present, Timely Language has connected to the new large-scale language model, and on this basis has developed a number of new functions including automatic generation of email content, multi-round dialogue, intelligent quality inspection, intelligent knowledge base, etc., completing the iteration and upgrade of the product.

In May this year , Timely Language released a new version combined with ChatGPT. The above new features will be gradually launched in the Timely Language customer service workstation. Bai Tao said that along with the launch of the knowledge base function, the company also supports the customization of customer service robots for cross-border sellers.

"In the past, the AI ​​models used mainly recognized user intent, determined what the user's question was, and then responded with a preset answer. Currently, our model has an accuracy rate of over 99% in language and emotion recognition, and an accuracy rate of over 95% in intent recognition," Bai Tao told 36Kr. "The new large language model solves the problem of rigid and inflexible intelligent answers in the past. At the same time, in order to ensure the accuracy and timeliness of the answers, we currently mainly use embedding to access model capabilities, and support two modes of AI automatic reply and human-machine collaboration in specific applications."

In terms of the market, Bai Tao introduced that in 2022, the number of customers and revenue of Timely Language both nearly doubled, and achieved a customer repurchase rate of nearly 100% despite price increases. At present, the company has established cooperative relationships with many large and medium-sized independent station groups, multi-platform sellers and DTC brand stations, and has served more than 100 customers in total, covering multiple categories such as clothing, 3C, and furniture. Timely Language's next plan is to achieve at least a 4-fold increase in the number of customers this year.

In general, the cross-border e-commerce intelligent customer service track still has a large market growth space in the future, but the competition is also relatively fierce. According to the data recently released by the Ministry of Commerce, there are more than 100,000 cross-border e-commerce entities in China, and more than 200,000 independent websites have been built. According to iResearch's 2022 estimate, the penetration rate of independent website marketing SaaS will reach 60%, and the penetration rate of platform sellers will be about 20%.

It should be noted that in addition to domestic and overseas companies of the same type as TimelyTalk, such as Leyan, Wisdom Tooth, Zendesk, and Freshdesk, participants in the track also include cross-border ERP SaaS whose product functions cover intelligent customer service. In addition, AI intelligent customer service companies that are currently focusing on the domestic market may also start to expand overseas markets at any time.

In this regard, Bai Tao said: "On the one hand, due to factors such as language, time difference, and culture, overseas companies have a strong demand for intelligent customer service that can provide 7*24 hours service and quick response, but the current layout of AI intelligent customer service in the field of cross-border overseas is still shallow. At the same time, the domestic players in the existing track focus on different segments and key markets. We mainly focus on the European and American markets, and other manufacturers may be more oriented to the Southeast Asian market at this stage."

In Bai Tao's view, the application scenarios and dialogue contexts of intelligent customer service in different markets are quite different, and scenario understanding, demand mining, corpus accumulation and model optimization all require a lot of time and effort. He told 36Kr: "We started paying attention to the cross-border market in 2019, and officially established the company in 2020, focusing on cross-border e-commerce. Our AI team has been exploring the field of natural language processing (NLP) for many years, and has gone through a long period of product co-creation with customers. This process requires solving many details, including support for multiple languages ​​and slang. Looking back, we have a certain first-mover advantage, and we do not rule out serving overseas brand sellers in the future."

At the same time, Bai Tao said that after this round of financing, the company will continue to polish its products. First, it will cover more seller categories and establish corresponding knowledge graphs; second, as the number of customers served increases, it will continue to optimize product functions to ensure product performance while improving network deployment efficiency; third, it will optimize AI models and further enhance the depth of integration between models and scenarios and products, and explore new functions that can create higher value for customers from the perspective of helping customers improve work efficiency and conversion rates, and reduce complaint rates and return rates.

In terms of payment model, after releasing the new version, JiShiYu will consider adding payment methods based on calling model API or traffic usage on the basis of the original subscription fee paid according to the platform account.

In terms of the team, Bai Tao, the founder and CEO of Shijiyu, graduated from the Automation Department of Tsinghua University. He has worked for China Mobile and UFIDA. He is the co-founder of Taiying Technology and has more than 20 years of experience in the construction and operation management of global CRM platforms. Feng Yan, the co-founder and CTO, is Bai Tao's college roommate. He has worked for China Mobile and is the co-founder of Wowotuan. He has 20 years of experience in product and R&D management in the Internet industry. AI Director Zhang Chaolin has more than 10 years of experience in AI scientific research and team management. He has worked for Century Dingli, ABC Technology, and ByteDance. Other core members of the team have experience in large cross-border e-commerce and Internet technology companies such as Dunhuang.com, Amazon, and SHOPLINE.

Investor Views

Mr. Meng Yaxi of Yuandu Capital said: "We have been paying close attention to AI and the overseas market, especially the commercial application of large models. The core members of the Timely Language team are all from Tsinghua University. They have profound industry experience and have conducted many years of exploration and practice in multilingual NLP models. They have achieved significant results in commercial applications and can well empower overseas companies, especially cross-border e-commerce companies, to help them achieve cost-saving and refined operations. Therefore, we are very confident in supporting and accompanying the Timely Language team to grow together. As an absolute leader in the overseas intelligent customer service market, we look forward to faster and better development of Timely Language."

Timely Words

New version of cross-border AI customer service