A technology start-up led by Hongkongers including a former Goldman Sachs analyst aims to help finance professionals and investors globally make better investment decisions with its artificial-intelligence (AI) chatbot that can perform rounds of “thinking and reasoning”.
Arbor’s AI chatbot ArborChat leverages large language models (LLMs) to analyse text and extract investment ideas to enhance fundamental stock analysis, the company said. About 100 firms have used it in trials, and a few have paid for the system, which was launched two months ago, according to Cheney Cheng, Arbor co-founder and CEO. The goal is to grow the client base to 100,000 in three years, targeting the global fund industry, he said.
“We believe this wave of AI is unlocking a big blue ocean in the investment area given the ability of AI to understand and analyse tasks efficiently beyond human capability,” Cheng said. “There will be a lot of uncaptured, profitable ideas.”
Investment has traditionally been “the crystal of humanity’s wisdom”, given the amount of information and the complex analysis skills required, making it one of the hardest areas for AI to conquer, said Cheng, a former banker in mergers and acquisitions who started his career in fundamental equity investment at Goldman Sachs Asset Management.
As the technology evolves, AI can now help with more sophisticated tasks.
For example, when asked how US president-elect Donald Trump’s policies will impact companies, or a list of thousands of stocks, ArborChat would go through various steps to break down the question and collect appropriate data before providing answers backed by databases within a few minutes.
The process is what the firm calls a “ThoughtTree” approach, emulating how humans would answer a question. ArborChat typically performs two rounds of questioning and three layers of analysis, and the company plans to increase this as the chip industry matures and computational power increases, Cheng said.
“To answer an investment question – or basically to answer anything that requires some thinking – we need multiple steps and to gather new information in each of the steps,” he said. The basic AI models lack that ability, as “they do not search and follow up”, Cheng added.
ChatGPT creator OpenAI and a slew of Chinese technology firms have rushed to unveil advanced AI models that claim to have reasoning skills in recent months.
Arbor’s AI tool is more tailored to the investment community and will be invaluable amid market uncertainties, Cheng said.
“The market is going to be very unpredictable, with a lot of major events that could impact it,” he said. “This kind of AI not only saves [analysts] a lot of time but also enables them to have a real-time response no matter what happens in the market.”
A portfolio adjusted according to ArborChat’s buy/sell/hold signals delivered a 41.75 per cent profit from October 2023 to September 2024, higher than the 34.25 per cent return in a benchmark, the Bloomberg 100 Index, Arbor said in November.
The AI chatbot currently covers US, Chinese and Japanese stocks, cryptocurrencies and foreign exchange, with plans to expand into other asset classes, including bonds and funds.
Last week, Arbor launched a feature that allows users to upload private documents. Fund managers can upload investment memos and private equity investors can upload deal documents to use the AI chatbot as an investment due diligence tool in a secured environment.
Established in 2018 by Cheng and his team, Arbor has provided AI solutions to banks and fund houses. It is understood that Hong Kong billionaire Richard Li Tzar-kai is a shareholder.
The London-headquartered firm is open to potential initial public offering “in any exchanges we can explore in the future”, Cheng said.
