Bloomberg introduced BQuant Assistant on October 28, 2026, a domain-grounded conversational AI coding assistant built into BQuant Desktop and BQuant Enterprise. It gives financial professionals who are Bloomberg Terminal subscribers the ability to use natural-language prompts to write code, build dynamic applications and accelerate financial data analysis workflows, inside the same Python environment where they already build investment strategies and dashboards.
Coding With Bloomberg Data, Not Generic Code
The key claim is grounding. A general-purpose coding assistant can write Python, but it does not know Bloomberg’s datasets, its query language or its libraries. BQuant Assistant, Bloomberg says, can answer questions about BQuant, generate Python code, help users query Bloomberg’s datasets through Bloomberg Query Language, help debug and iterate on their work, and build code for data visualisations. It can also search the BQuant Help Center and draw from example notebooks, libraries and how-to guides when answering.
That grounding is meant to produce context-aware responses about a user’s own projects, workflows and code, rather than plausible-looking code that fails against real financial data. In quantitative finance, that distinction is everything: a query that returns the wrong field, the wrong date range or the wrong corporate action adjustment can quietly corrupt an entire analysis, and the error may only surface after a decision has been made.
Users can generate BQL and PyBQL queries to retrieve and analyse Bloomberg financial data spanning equities, funds, fixed income, foreign exchange, commodities, indices, derivatives and other asset classes, across more than 22,000 fields, including reference, pricing, returns, derived analytics, fundamentals, valuations, sustainable finance data, risk metrics, estimates and economic data. They can also create code for screens and universes across equities, fixed income, foreign exchange, commodities, indices, exchange-traded funds and other asset classes.
Control Stays With The User
Bloomberg has built approval into the assistant’s actions. It can respond with code suggestions and, subject to the user’s approval, insert and run code directly in an active project within the secure BQuant sandbox, so users can move from question to result without switching context. Users retain control, Bloomberg says, because the assistant identifies the action it proposes and requests approval before reading from or writing to a notebook.
That matters in a regulated, high-stakes environment. A coding assistant that silently edits a notebook or runs code against live data would be a compliance problem, not a convenience. By naming each proposed action and asking first, BQuant Assistant keeps the human as the author of record, even as it speeds up the drafting.
The assistant can also answer questions about Bloomberg libraries and workflows, discuss financial concepts and generate code to implement them, recommend visualisations using libraries such as Plotly and Jupyter Widgets, and explain selected code to help users build Python knowledge while working.
What Bloomberg’s CTO Says It Is For
Shawn Edwards, Bloomberg’s chief technology officer, said financial professionals are working with increasingly large and complex datasets, but turning an investment idea into a repeatable analysis can still require significant time and specialised coding knowledge. BQuant Assistant, he said, brings AI-driven coding informed by Bloomberg’s understanding of financial data and quantitative workflows directly into the workflows clients already use, helping every BQuant user — from a portfolio manager sketching a first query to a quant engineering a multi-asset signal — move from a natural-language question to working code, data retrieval and visualisation.
The rollout is staged. BQuant Assistant will be generally available in BQuant Enterprise, and in limited beta in BQuant Desktop, at the end of October 2026. Desktop is included with a Bloomberg Anywhere subscription and gives Terminal users programmatic access in a secure sandbox, while Enterprise is a fully managed, cloud-based platform with large-scale compute and collaborative development.
The launch follows other recent Bloomberg AI announcements, including ASKB, the Terminal’s conversational AI interface, and an enterprise Model Context Protocol layer for Data License Plus, its next-generation Data License offering. Together, they show Bloomberg building AI into each layer of its stack — terminal, data licence and quant platform — rather than betting on a single chatbot.
Why It Matters
This matters because quantitative finance has always been limited by coding skill, not by ideas. A portfolio manager with a good hypothesis but weak Python skills has historically had to wait for a quant to implement it, and that delay kills many ideas before they are tested. If a domain-grounded assistant can reliably turn a plain-language question into working BQL and Python, it shortens that loop from days to minutes, and lets more people test more ideas. The same democratisation is at work in developers pointing rather than typing their fixes and accountants working with AI on one platform: the tools that win are the ones that meet professionals inside the environment they already use.
The risk is the usual one for AI-generated code: it looks right, runs without error, and is subtly wrong about the data. Bloomberg’s grounding in its own datasets and documentation is its defence, but users will still need to check outputs against known results before trusting a new strategy built on assistant-generated queries.
Background
Bloomberg introduced BQuant Assistant on October 28, 2026, for BQuant Desktop and BQuant Enterprise, with general availability in Enterprise and limited beta in Desktop at the end of October 2026, as part of its wider AI programme, which Bloomberg says dates back to 2009.
Source: Bloomberg announcement
0 Comments