36Kr Exclusive | AI Phone Digital Employee Company "Whobot" Completes Tens of Millions of Yuan in Series A Financing with Human-like Conversations

Author | Zhong Yixuan

36Kr learned that the AI phone digital employee company "Whobot" (Hubot Artificial Intelligence) recently completed a Series A financing of tens of millions of yuan. This round of financing was led by GSR Ventures, and Xiangyang Capital served as the exclusive financial advisor. The funds obtained will be invested in key areas such as technology R & D and market expansion.

Whobot was founded in 2023. It is a digital employee usage platform that provides enterprises with intelligent one - stop solutions specifically for communication scenarios. Through self - developed core technologies such as intelligent large language models, voice interaction systems, and real - time operating systems, Whobot's AI digital employees aim to replace manual labor with AI, break the barriers between dialogue and execution, and bring improved conversion efficiency to enterprises. Whobot's product has been officially launched for only one year. The current team consists of 13 people, and it has signed contracts with more than a hundred customers. The renewal rate of formal customers reaches 100%.

Different from traditional telemarketing platforms or SaaS platforms, Whobot can reduce the cost of manual calls, is not limited by human work and rest schedules, and based on the service phones provided by enterprises, it offers a product system including five major aspects: intention mining, lead WeChat addition, product sales, hotline answering, and service communication.

Founder Dong Lianping summarized Whobot's greatest feature to 36Kr in one sentence, which is "anthropomorphism".

Precision and speed are the characteristics of AI, but language expression is a complex and fluid art, which can be affected by word choice, tone, and even some hard - to - capture feelings. "Whobot's AI phone digital employees will try their best to imitate the scenario of a real person having a conversation while operating," Dong Lianping said.

He simulated such a scenario on - site -

A user claimed to be busy in a meeting and hurriedly hung up the customer service call.

If it were an inadequately trained AI customer service, it would call back a few minutes later and repeat the description of the demand in a rigid language, which would most likely arouse the user's disgust.

But Whobot will remember the user's expression, choose to call back an hour later, and after connecting, use a transitional language to make the context more natural, such as "I'm XXX who called you just now."

Image source: Official

Mechanical response speed, single tone processing, and fixed and cold scripts are all characteristics that humans use to identify AI today. "So when communicating with human voices, the physical experience is very important, because once the other party finds out that you are an AI, they will instantly feel disrespected," Dong Lianping said.

Therefore, through the large - model dialogue engine and multi - modal decision - making system, while Whobot's AI digital employees can handle long conversations with a 300% improvement, they also have the habits or skills of human conversations, such as actively guiding the conversation, pausing appropriately, adding modal particles, and simulating real - human tones. In addition, when the user suddenly interrupts or pauses in thinking, the AI digital employees can also recognize that the other party's expression is incomplete and will not show obvious AI characteristics such as "interrupting", "talking to themselves", or "taking things out of context".

Dong Lianping told 36Kr that according to Whobot's internal statistics, 65% of users have regarded Whobot's AI digital employees as real - life customer service representatives.

These "anthropomorphic" features all come from the real corpus of top - notch salespeople. Currently, Whobot has accumulated hundreds of business corpora. With a large amount of corpus feeding, AI digital employees can quickly learn business and identify scenarios.

Based on the corpus, Whobot can achieve full - industry coverage of any scenario with a phone - related position, that is, "as long as there is a phone - related position, Whobot can replace it". Currently, Whobot has completed deliveries in the education industry, the automotive industry, various customer service fields, and operator SP businesses, and has achieved cost reduction and conversion rate improvement for the above - mentioned businesses. Taking the base of 1000 conversions in the operator SP business as an example, the cost of real - life customer service is about 300,000 yuan, while the cost of Whobot is only 30,000 yuan.

In addition, Whobot has also broken the barriers between dialogue and execution, realizing real - time cross - system operations and a closed - loop delivery. Different from the previous AI phone robots that could only perform simple notifications and screenings, Whobot's AI digital employees can do things "while talking" just like real people.

Taking educational institutions as an example, the private domain is a necessary way for them to achieve conversions. Traditional AI phone robots can only reach users via text messages after obtaining customer intentions, guiding users to add service personnel on WeChat by themselves, which will inevitably lead to customer loss. However, Whobot's AI digital employees can interact with users in real - time. In the scenario of selling courses and adding WeChat contacts, they can perform operations on the computer just like real people, achieving both positive addition (service personnel adding users) and reverse addition (guiding users to add service personnel).

Whobot once counted that in the scenario of adding WeChat contacts for a 19.9 - yuan trial course, the conversion rate of real - life customer service is about 87%, while Whobot's AI digital employees can achieve a conversion rate close to 95%.

Dong Lianping also introduced to 36Kr that Whobot's AI digital employees can not only be used to improve conversion rates but also realize the "anti - cheating" function in reverse. For example, in the lead industry, Whobot can identify missed conversion cases by quickly tracking users' transaction trajectories.

Dong Lianping, the founder of Whobot

In terms of the R & D team, founder Dong Lianping has served as a senior executive at Zuoyebang and a senior technical manager at Baidu's large - scale search. He has 14 years of experience in the Internet industry and is a serial entrepreneur. CTO Liang Bin has served as a senior technical expert (P8) at Alibaba Cloud and a senior R & D engineer (T7) at Baidu. He has nearly 20 years of experience in the technology field and management. The team members are good at using technology to solve business problems.

"Whobot only provides productivity and does not determine production relations," Dong Lianping said. He believes that the value of Whobot does not lie in replacing humans but in redefining the service boundaries. When AI undertakes standardized and high - concurrency primary services, humans will eventually be able to focus on more creative work one day.