Toss Securities wants investors to open its app for investment information even when they are not trading stocks. Rather than building around the brokerage's own priorities, the company designed its artificial intelligence services to follow the path novice investors actually take — starting with individual stocks, then widening to news events and the broader market.
"We developed our AI services around how investors actually move through the app," Wang Hyun-min, product owner of Toss Securities' AI Tribe, said in an interview with The Seoul Economic Daily on the 21st. Wang leads the AI Tribe after stints at Line Plus, Grab, Facebook and Coupang Play. Launched in January, the AI Tribe is dedicated to customer-facing AI services at Toss Securities, and has produced AI Signal, Earnings Call and Real-Time Issues. This year the team added former analysts to deepen its financial analysis expertise.
Changes to the services began with investors' questions. To answer users asking why a stock they hold was falling even without any bad news, the company expanded from AI Signal, which focused on individual stocks, to Real-Time Issues, which compiles major news. On the 14th of this month, it added a Market tab showing index and sector moves along with buying and selling by foreign and institutional investors. "What people want to know these days is as much about why the market is shifting as it is about why a single stock rises or falls," Wang said.
What has changed is the scope of the information. Existing stock rankings were based on trading value and volume among Toss Securities users, while the new Market tab reflects trading across the entire market. AI Signal now also incorporates supply-and-demand data, explaining the flow of foreign and institutional trading even when there is no news or regulatory filing. Cumulative users of Real-Time Issues reached 3.12 million at the end of September, and AI Signal averaged more than 1.5 million monthly users from July through September.
Wang pays more attention to how often investors return for information than to trading figures. His view is that as AI features at brokerages converge, keeping customers on the platform longer will determine competitiveness. "We don't look at trading metrics," he said. "How often people use the information and how often they come back — those are the key results." The goal of becoming "an Investing.com that beginners can use" rests on the same logic: maintaining contact with investors even in moments when they are not trading.
Accuracy is the prerequisite for a service people return to. The AI Tribe has built a reference data set it calls the "golden set," which categorizes factors that move share prices such as supply and demand, regulatory filings and share buybacks, and uses past cases to evaluate the AI's explanations. Three staff members review the output continuously and feed errors back into the golden set. In one case, an explanation for a share price gain remained on screen after the stock had already reversed and fallen, confusing investors, prompting the company to change how explanations are displayed.
The next task is making the information more sophisticated. Toss Securities is preparing to launch an AI service this year that combines corporate earnings data with earnings calls to help beginners understand results more easily. Wang expects "hyper-personalization" to arrive within one to two years, in which the order of information shown and the way the app is used will differ according to a user's watchlist and trading habits. He was more cautious about AI taking over buy and sell decisions. "In line with what investors are asking for, we are also considering strengthening the heat map in the Market tab and linking rankings of posts on X, formerly Twitter, to market news," he said.