Cala Terminal
What Bloomberg actually announced
Bloomberg’s October 15 announcement describes general availability of its Terminal conversational AI, with expanded use of enabled IB chats, Worksheets and files. Bloomberg says source references accompany answers and private material follows sharing permissions. General availability does not mean free public access.
The ASKB product page also describes reusable workflows, saved lists, permissioned firm research and continuity between desktop and mobile. These are vendor-described capabilities. We have not tested their accuracy or availability under a particular contract.
Why persistent context changes the question
A one-off company summary answers what a model can retrieve now. A useful research process also asks what you believed last quarter, which assumptions changed, and which evidence contradicts your view. That makes the organization of notes and sources part of the product, rather than an attachment to a conversation.
Our interpretation is that financial AI evaluation now needs two separate tests. Can the system retrieve the right public fact? Can it apply the right private context without using material the reader is not allowed to see? Passing one does not establish the other.
A citation proves that a source exists only after you open it. Check whether the passage supports the exact claim, whether a table uses thousands or millions, and whether the reporting period matches the question. A source about revenue does not establish a statement about free cash flow.
Five checks before trusting an answer
Require the security, exchange and currency. A ticker alone can be ambiguous. Separate a share price, a depositary receipt, an index and a derivative.
Record the observation time, time zone and session. Ask whether the figure is a last trade, a quote, a previous close or an end-of-day value.
Follow citations to the actual document or data record. Save the filing date, period and relevant passage. An attractive summary cannot replace this step.
State the formula and the inputs. Keep reported, adjusted and estimated values separate. A correct formula applied to mismatched periods still produces a misleading answer.
Ask what is missing and what would change the conclusion. A credible response can say that a data point or permission is unavailable.
A small test you can repeat across tools
Choose a company and a completed reporting period. Obtain its filing from the company’s investor-relations page or SEC EDGAR, which provides free public access to filings. Ask each tool the same bounded question and keep its response with the source document.
For example, ask for revenue growth against the corresponding prior-year period, with reported figures and a formula. Check the values in the filing and calculate (current revenue ÷ prior-year revenue − 1) × 100. Use comparable periods, currency and restatements, with prior revenue greater than zero. Otherwise mark percentage growth as not applicable and explain why. This is an evaluation exercise, not a recommendation to buy the company.
Record the result as supported, unsupported or unavailable. Also record the time needed to verify it. Do not turn a small sample into an overall accuracy percentage, or compare live prices captured at different times as though they were synchronized.
For this company and reporting period, identify reported revenue and the comparable prior-year figure. Cite the original filing, show units and currency, and calculate year-over-year growth. Flag missing information.What this means when you try Cala
Cala is built for individual market research. We publish this checklist so you can assess it by the same standard as other tools. It is not a claim that Cala provides Bloomberg’s licensed datasets, institutional messaging, research entitlements or full workflow coverage.
Use one task and your own evidence when trying Cala. If the current product cannot retrieve a source or complete the task, treat that as a limit. A development plan is not a live feature. Consult the current access and subscription terms before subscribing.
AI output is research material, not a guaranteed result. The SEC, NASAA and FINRA warn against promises of AI-generated investment returns in their joint investor alert. Our articles do not provide personalized investment advice.
For a task-based buying framework, read how to compare Bloomberg, Cala and retail research tools.
Frequently asked questions
Is Bloomberg ASKB a free public chatbot?
No. Bloomberg describes ASKB as a conversational interface within the Bloomberg Terminal. General availability for financial professionals does not mean free public access. Confirm access and contract terms with Bloomberg.
Does a cited AI answer guarantee accuracy?
No. Open the cited source and verify the instrument, reporting period, units, timestamp and calculation. A source may be real while the answer misreads it.
Did Cala test ASKB in this article?
No. This article reviews public documentation as of October 19, 2026. It does not report an authenticated product test or a comparative accuracy benchmark.
Try one research question in Cala
Bring a company, a reporting period and a source you can check. Judge the answer by its evidence and the work it saves you.
Open CalaDownload apps