The inside story of Taiwan’s AI whizz

When Chih-Han Yu’s work on multi-agent artificial intelligence (AI) was nominated as the best doctoral thesis of the year in 2010, the rising star in AI was not content with his stellar achievements, which included an early prototype self-driving car that laid the foundation for Google’s self-driving car project.

“People knew we were publishing high-quality research, but back in my dorm, my room-mate and I were thinking that we’ve worked on all this coding, but we have never seen any algorithm that has really impacted the world and transformed how people live and how business is done,” said Yu, referring to his time at Harvard University.

The duo decided they should do something and started a company specialising in AI-powered game engines that mimic the actions of human gamers, based on Yu’s doctoral thesis. But that proved to be a mistake, said Yu, because there was no demand for the technology at the time.

Two years later, they pivoted the business that would later become Appier, a supplier of AI-based marketing technology that helps businesses improve customer engagement and drive sales at a time when interest in big data was growing. “Businesses have big data, but they don’t know how best to use it, so we started Appier to help companies make better decisions by utilising their data,” he said.

The early years were tough for the co-founders. Confronted with rising costs, they moved from Boston, where they started the company, back home to Taiwan. Since then, the company, which raised $80m in Series D funding in November 2019, has grown to about 500 employees across 15 cities in Asia and Europe.

Yu attributed the company’s success to its ability to solve a fundamental problem for businesses – customer engagement. “We’ve aligned our product lines across different stages of the customer journey – from acquiring them, retaining them, transacting with them and predicting their future behaviour,” he said.

It also helped that customer engagement – and more broadly, digital marketing – is an area with more structured data that can be harnessed to build a standardised AI model for a customer base. “Digital marketing is also at the forefront of the AI revolution because customer and data readiness are among the highest across domains,” said Yu.

That does not mean Appier’s algorithms cannot be applied to other domains. In fact, because its software enables businesses to automate decision-making, industries such as manufacturing can benefit from the same technology.

Noting that it is not a question of whether Appier’s technology will work for a given domain, Yu said it is about identifying business cases. “With our four product lines, we’re gradually expanding to adjacent fields to help more companies automate decision-making and become more efficient, beyond digital marketing and sales,” he said.

Appier is not the only player in the market selling AI technology to digital marketers. Customer relationship management (CRM) and e-commerce software suppliers are already baking AI capabilities into their platforms.

But unlike the platform players, said Yu, Appier’s offerings are its core products, while those available in CRM platforms are like the “gravy over the product”.

“I’m not saying which one’s better, but if a customer wants to forecast and take proactive action, they probably find our products more suitable,” he added. “But if they just want to store data and have some basic rule-based automation, they will probably find that other more traditional software can help them better.”

But knowing what a customer wants and will want in future is just the tip of the iceberg. Increasingly, businesses are seeing the need to orchestrate business processes between systems of engagement and the supply chain to fulfil customer demand and deliver a good customer experience.

Acknowledging this, Yu said: “That is a great potential and we continue to dig deeper and have even tighter integration with back-end systems so we can get real-time data to predict better.”

In explaining AI-driven decisions to users, which is crucial to foster trust in the technology, Yu said Appier’s data science platform that predicts the future actions of customers shows the most important variables in an AI model and why certain choices and decisions are made.