Bloomberg expands ASKB access with persistent files and chat search

Bloomberg made its Terminal-based ASKB interface generally available on 15 October, adding wider support for users’ conversations, security lists and files. The change brings private working material into the same research process as market data.[1]

For a brokerage research team, the commercial value lies in reducing the work between receiving information and testing an investment view. A system that can revisit an earlier thesis may make new information easier to assess. Its value still depends on whether the analyst can distinguish a recent observation from an assumption that has gone stale.

The new inputs change the research task

A shared folder introduces a different editorial problem from a public company filing. It may contain a draft forecast, a superseded model and a final investment committee paper about the same business. Access permission answers who may read them. An analyst also needs to know which document carries the current view and which assumptions have changed.

The October rollout builds on an existing research network

ASKB already had external research inputs during its beta. Third Bridge announced its integration on 16 April, making expert-interview content available alongside Bloomberg’s other research tools. It described access to a library of more than 100,000 transcripts and attribution back to the underlying content. That earlier integration helps distinguish October’s availability and private-content expansion from the original introduction of research search.[3]

The product page describes scheduled workflows, reusable templates and workflows that colleagues can share. It also supports continuing a desktop conversation on the Bloomberg Professional mobile app. Those features could make a research process repeatable across a team, provided its inputs and instructions remain appropriate for the question being asked.[2]

Repetition can reduce manual setup, but it also repeats an error in the setup. A narrow watchlist can leave out a newly relevant competitor; an old instruction can keep asking the wrong comparison after a company changes its reporting segments. The operating task shifts towards maintaining the research process as well as reading each answer.

A supplier’s controls sit inside the firm’s responsibilities

FINRA’s June 2024 notice says its existing rules apply when member firms use generative AI, including features embedded in third-party products. It calls for evaluating tools before deployment. Where AI forms part of a supervisory system, the notice identifies model governance, data privacy, integrity, reliability and accuracy as relevant considerations. The obligations depend on the business use.[4]

That distinction matters when a research summary becomes something else: a message to a client, an input to a recommendation or evidence in a supervisory review. Each use has a different audience and consequence. A citation makes checking easier, but the existence of a cited passage does not by itself demonstrate that the conclusion follows from it.

Bloomberg’s product description emphasises source attribution and firm-level permissions. Those controls provide starting points for checking what an answer used and who can see it.[2]

The rollout broadens the work that can begin inside the Terminal. The useful measure is completed, reviewable research: how much relevant evidence a team can examine, how reliably it catches contradictory information, and how clearly it records why its view changed.

Report an error or submit an update