Bloomberg is rolling out a chatbot-style interface called ASKB for its Terminal platform, allowing finance professionals to query vast datasets using natural language rather than manually navigating the software’s traditional interface. As of late April 2026, the ASKB beta is available to roughly a third of the Terminal’s 375,000 users, with no full release date yet announced.
The development was detailed in an interview with Bloomberg’s chief technology officer, Shawn Edwards, conducted at the company’s London headquarters in early April 2026. ASKB is built on a combination of multiple language models and is designed to help users synthesize large volumes of financial and alternative data — including earnings, asset prices, weather forecasts, shipping logs, factory locations, consumer spending patterns, and private loans.
Edwards described the core problem the tool aims to solve: as Bloomberg has continued expanding its dataset over the years, locating relevant information has become increasingly difficult. “It has become more and more untenable,” he said. “You miss things, or it takes too long.”
Rather than querying specific data points, users can pose broad investment questions through ASKB. Edwards offered an example: “How is the war in Iran and a change in oil prices going to affect my portfolio?” The system is intended to synthesize an answer drawing on multiple data dimensions.
Bloomberg also pitches ASKB as having agentic capabilities. Edwards described a workflow feature that allows analysts to build templates — for example, automatically preparing research packages ahead of quarterly earnings calls — and schedule or trigger them based on specific market conditions.
Edwards was measured about the tool’s limits, noting it would not elevate the quality of an analyst’s underlying thinking. “These tools are not magical,” he said. “If you’re a mediocre analyst, they’ll be 10 mediocre ideas.” For experienced professionals, however, ASKB could allow deeper research across more ideas in less time, he suggested.
Source: WIRED