Bloomberg rebuilds Terminal around AI: the ASKB chat will become its main interface

Bloomberg is preparing the most visible update to Terminal in years: the ASKB chat interface is being built into the legendary platform for traders and analysts. The idea is that users no longer have to work through layers of commands, screens and notes taped to a monitor. Instead, they ask a question in plain language and immediately get a compiled answer about markets, companies and documents.

Why Terminal is changing

Bloomberg Terminal's strength has always been the density of its data and functions, but over time that became a problem. For decades the platform accumulated new layers of information: beyond quotes and financial statements, it now carries weather forecasts, logistics data, factory information, consumer spending and private credit. As a result, even experienced users find it harder to quickly locate the signal they need in this mass of data.

According to Bloomberg CTO Shawn Edwards, the main job of generative AI here is not to "think" for the analyst but to find key insights faster and build a coherent picture around an investment idea.

The ASKB beta is currently open to roughly a third of Terminal's 375 thousand users, but Bloomberg has not yet named a date for the full release.

Most importantly, ASKB is conceived not as a side feature but as the new main way into the system. Bloomberg says outright that the graphical interface is not going away, but most workflows will start with a conversation. For Terminal this is almost a cultural upheaval: instead of memorizing commands and routes across screens, the user states a thesis and gets a compiled answer.

"This will be the new

Terminal: most interactions will start with ASKB."

How ASKB works

On the surface ASKB looks like a chat, but under the hood it is not a single all-purpose bot. Bloomberg describes the system as a network of AI agents that work in parallel across data, news, research, documents and analytics, then assemble a contextual answer. For finance professionals this matters: you can ask a question not in the language of tickers and commands but in the language of a hypothesis, for example how a geopolitical conflict and oil price moves could affect a portfolio.

In practice, Bloomberg pitches ASKB as an accelerator for research, especially in periods like earnings season, when analysts must review a mass of documents in a short time and compare companies with their competitors.

• Natural-language search across data and documents
• Summaries of bullish and bearish scenarios for a company
• Workflow templates for earnings prep and recurring tasks
• Transparent links to source passages and verifiable quotes
• BQL code output for continuing analysis in Excel and BQuant

Edwards stresses that such tools do not automatically make an average analyst great. But they let experts test more ideas in the same time: not one hypothesis a day, but ten. In other words, the competitive edge shifts away from whoever best remembers Terminal's internal functions toward whoever asks stronger questions and interprets the answer faster.

Risks and resistance

For financial software the main risk is obvious: a confidently worded error can cost too much. That is why Bloomberg emphasizes safeguards rather than a flashy chat. According to the company, the system checks whether the facts in a summary appear in the source paragraphs, separately tracks the semantic inversions that language models like to produce, and validates citations. The approach is described as conservative: better to under-answer than to deliver a convincing hallucination.

A further principle follows: ASKB should lead the user to sources, not hide them behind a polished paragraph. This is an important difference from mass-market AI chats. In investing, what counts is not only the speed of an answer but also the ability to quickly open the original document, check the wording and understand exactly what a conclusion is based on. Bloomberg also promises transparent attribution and, in some scenarios, shows fragments of the original text next to the answer.

There is an organizational question too. If AI starts automating the routine that junior analysts used to do, the logic of training inside the industry changes. Edwards admits the market has not yet found a clear answer on how to develop newcomers if a significant share of the "grunt" research work goes to agents. Bloomberg is also clearly preparing for discontent from the old guard, for whom knowing Terminal's functions was part of their professional identity.

What this means

Bloomberg shows that generative AI in finance is moving out of the experimental phase and becoming the interface to a mission-critical system. If ASKB takes hold, the market gets a new standard: simply providing access to data is not enough. Platforms must be able to assemble conclusions quickly, show where they come from and automate research routine without losing control.