Most of us have spoken to voice assistants, whether it is Siri, Alexa, or Google Assistant. Some have even dealt with customer service agents powered by conversational artificial intelligence applications.

AI talkbots have been adopted by enterprises around the world to provide customer service at scale. The ASEAN market, however, represents a major challenge for AI conversational firms due to the many languages and dialects spoken in the region.

Jennifer Zhang, CEO and co-founder of Wiz.AI, saw an opportunity. “When we started in early 2019, there was nobody actually bringing voice AI products to Southeast Asia,” she told KrASIA. “Doing this in Southeast Asia is very challenging, but we want to target this market as our starting point to create our competitive technology advantage.”

Singapore-based startup Wiz.AI develops conversational AI technology. Its solutions serve enterprises of all scales in the region. The firm, which has offices in Nanjing and Jakarta, closed a USD 6 million pre-Series A round in May last year, led by GGV Capital. Wavemaker Partners, ZWC Partners, Insignia Ventures, and the Orion Fund managed by K3 Ventures were also part of the investment.

Conversational AI is a “programmatic and intelligent way of offering a conversational experience to mimic conversations with real people, through digital and telecommunication technologies,” according to Deloitte Digital. The firm expects the global conversational AI market to reach USD 17.64 billion by 2024, carrying a compound annual growth rate of 30.2%.

Jennifer Zhang, CEO and cofounder of Wiz.AI. Photo courtesy of Jennifer Zhang.

Automated customer voice services  

To foster a natural, smooth conversation flow between an AI chatbot and a human user, the first step is to recognize the speaker’s speech pattern. That means understanding her language or dialect, slang, as well as inconsistencies and interruptions in lines, Zhang explained. The process involves automatic speech recognition to convert spoken words into text, running alongside natural language processing to interpret the meaning of that text.

The next step is for the chatbot to provide a valid answer. For this, Wiz.AI has a dedicated team that performs “conversational design” to generate dialogue mimicking realistic human-to-human interactions. The team has developed voices and tones for different scenarios to provide “a better engagement feeling,” Zhang said.

For instance, a middle-aged woman’s voice—gentle and comforting—is preferred in a healthcare scenario, said Zhang. In another common situation, talkbots are able to answer customers’ questions such as How much should I pay? or Where to pay? They can even “bargain” with the customer, steering them to the minimum acceptable charge. If a speaker does not speak clearly, the chatbot will issue queries like Do you mean this? or Can you repeat yourself? And, of course, it asks for and logs all the necessary information from customers.

Wiz.AI’s talkbot automates outbound calls for enterprises to reach their customers. Graphic courtesy of Wiz.AI.

Wiz.AI has developed talkbots that are able to communicate with customers in several languages, including English, Mandarin, and Indonesian. They also cover informal forms such as Singlish, a blend of Singaporean slang and English, as well as Manglish, an informal variant of Malaysian English. At present, the company counts over 30 clients from various industries, including healthcare, insurance, banking, telecommunications, e-commerce, and the government.

AI voice assistants are optimized to answer particular clients’ demands, Zhang said. For example, for healthcare clients, Wiz.AI’s talkbots proactively reach out to patients, track their symptoms, and help them arrange health visit appointments. For another client, Singaporean virtual telco Zero1, talkbots call users to offer promotions and address their queries.

Wiz.AI conducted a survey and found that 97% of medical patients are comfortable with receiving calls that utilize talkbots, Zhang said to KrASIA. Rather than sending emails or using online chat clients, most people still prefer to make phone calls for complex interactions compared. 

Even so, the company also offers automated chat apps for their customers. On top of that, Wiz.AI leverages customer relationship management software for clients to track and manage data about their customers’ needs.

A virtual assistant developed by Wiz.AI checks in with patients and arranges hospital appointments. Graphic courtesy of Wiz.AI.

‘Last-mile delivery’

Wiz.AI has its work cut out for itself.

When asked about running up against Big Tech players like Apple, Amazon, Google, and Microsoft that make similar applications, Zhang responded with a metaphor. “They’re making the freeway, but we’re handling the last-mile delivery.” Wiz.AI positions itself as a service provider in “the untapped and challenging Southeast Asian market, which might not be the priority of big firms.”

Last year, Wiz.AI’s talkbots found rapid, immediate use cases. The COVID-19 pandemic suspended the operations of many call centers around the world. Budget slashes meant the workers who staffed these facilities were laid off or put on furlough. That’s when Wiz.AI’s products filled in, replacing fleets of human operators.

Even though there were challenges in integrating Wiz.AI’s talkbots with the existing infrastructure of some enterprises, the startup used these opportunities to test out new application scenarios and fine-tune their products, connecting their talkbots with many people around the world.

This article is part of KrASIA’s “Startup Stories” series, where the writers of KrASIA speak with founders of tech companies in South and Southeast Asia.