S&P Global connects Capital IQ Pro AI answers to financial source data

S&P Global’s AI proposition combines company and market information with tools that help professionals interrogate it. Capital IQ Pro brings that approach into a research workstation through ChatIQ and Document Intelligence. The practical attraction is a shorter path from a question to an inspectable source, especially when an analyst needs to connect financial figures with what a company has said in its filings and calls.

01 / ProductAn AI interface over a specialist research workstation

The Capital IQ Pro product page identifies ChatIQ, Document Intelligence and Chart Explainer among its AI capabilities. The product combines financial information, company research and analytical tools. This blueprint uses S&P Global as the company identity and concentrates on that Market Intelligence offer, rather than treating the group’s ratings, indices and other businesses as a single application.

The ChatIQ launch announcement identifies Kensho as S&P Global’s AI innovation hub. It describes ChatIQ as a natural-language research interface with traceability to source data, while Document Intelligence brings search, document interaction and summarization together. Kensho is therefore part of the company’s offer here, not a separate new company listing.

The key distinction is between a generated explanation and the information behind it. An analyst might use the same workspace to explore a question, inspect the filing and examine a standardized financial item. Those steps should remain distinguishable in the resulting work. A fluent explanation does not by itself resolve whether the source, accounting basis and reporting period are appropriate.

02 / AudienceFor company research that mixes numbers and narrative

A strategy analyst might compare how several competitors discuss a common cost pressure. A transaction team might look for changes in a target’s operating model. An investor-relations team might examine the questions raised on peer earnings calls. These jobs require both structured information and careful reading of company language, which makes the workstation a sensible product to evaluate.

S&P Global’s professional-services page describes document research, peer analysis and extraction of data tables into Excel. Those are useful capabilities to demonstrate on a representative assignment. They do not establish that every type of broker research or every data field is included in a particular customer’s subscription.

The AlphaSense blueprint is relevant when the main task is finding and synthesizing business documents. The Databricks blueprint helps frame a different option: building analysis around an organization’s own data platform. Compare the work the team wants to own. A specialist workstation can reduce assembly work, while a custom environment may be appropriate when proprietary data and bespoke calculations dominate.

03 / WorkflowA proposed comparison of company disclosures

For a bounded evaluation, select a small peer group and a past reporting period. Choose a question such as how companies describe a particular operating constraint, without asking the system to select investments or predict returns. Include a company that changed its segment presentation so that the exercise tests comparability rather than only document retrieval. Sequenced has not run this workflow.

First define the peer group and explain why each company belongs in it. A shared industry label is not enough when businesses have different product mixes, geographic exposure or reporting structures. Ask the assistant to identify potential mismatches, then have the analyst approve the list. Retain the excluded candidates and the reason for exclusion in the working notes.

Use ChatIQ to develop an initial set of themes and follow its source references. Distinguish management’s explanation from the analyst’s interpretation. For instance, a statement that demand softened in one region should not become a claim that the entire company experienced the same change. Check the relevant document and reporting period before adding the observation to the comparison.

Use Document Intelligence to inspect the filings, transcripts and other licensed materials behind the themes. Extract only the tables necessary for the research question, then verify units, signs and footnotes. A table with clean columns can still contain inconsistent currencies or a mixture of annual and quarterly figures. The analytical record should state how those differences were handled.

Create a concise evidence matrix: company, period, relevant passage, comparable metric and unresolved question. Ask the assistant to identify contradictions between the numerical trend and the narrative. Treat any suggested explanation as a hypothesis until the supporting disclosure is found. If the company gives no explanation, preserve that absence rather than asking the model to complete the story.

End with an internal briefing that separates observed facts, management statements and the team’s analysis. Have another analyst reconstruct several conclusions from the linked sources. Measure omitted disclosures, entity mistakes, inconsistent periods and time needed to verify the packet. The test should reward a useful research trail and clearer judgment, not merely the number of generated paragraphs.

04 / PricingBuy the data and workflow scope the team actually needs

The product page offers request-access and sales-contact routes. It does not publish a universal Capital IQ Pro tariff in the reviewed material. Obtain an offer covering the required users, data collections and AI functions. A demonstration should use the same entitlements as the proposed package so that the evaluation reflects what the team can actually access.

The buy-side AI page describes AI-ready data delivery and MCP connections as well as workstation tools. Those routes matter for firms building their own applications, but a desktop subscription should not be assumed to include every API, redistribution or external-model use. Scope a custom data project explicitly with the relevant commercial team.

For an existing customer, ask which functions are already enabled and what requires a change to the agreement. Also establish how licensed research and generated outputs may be shared with colleagues or clients. The commercial comparison should reflect the whole research job, including source access and review time, rather than an informal price comparison between differently configured accounts.

05 / DistinctionsStructured data can challenge the narrative

The useful distinction is the proximity of structured company information to textual evidence. An analyst can ask whether a narrative theme is consistent with reported figures and then inspect the relevant material. This is especially valuable when a company’s language remains reassuring while its reporting shows a shift that deserves closer investigation. The assistant should make that comparison easier to examine.

S&P Global’s corporate workflow page discusses analysis of financial documents, market impacts and peer disclosures. Its examples help frame potential tasks, but they are vendor descriptions rather than measured outcomes for a prospective customer. A pilot should isolate the particular document types and questions that the team handles repeatedly.

Traceability also changes the nature of review. Instead of debating whether a generated paragraph sounds plausible, colleagues can inspect the passage or data item supporting it. That only works when links remain usable for the reviewers and the cited material matches the claim. Source availability for the author and source availability for the wider team should both be checked.

06 / QuestionsTest the source, the time basis and the delivery route

One question is whether the team can distinguish current information from a point-in-time view. A historical comparison can be distorted if later corrections or disclosures enter the answer without being marked. Ask the vendor to demonstrate how the chosen workflow handles publication dates, revisions and the period specified in the question. Preserve the basis of the internal analysis accordingly.

Another question is how the system treats ambiguous financial definitions. Revenue, operating income and adjusted measures can have different scope across companies. A natural-language prompt may conceal a choice the analyst would normally make explicitly. Request the definition and source for each metric, and reproduce a sample of the calculations independently before accepting the comparison.

A third question concerns moving from workstation research to a custom application. The buy-side material shows that several delivery paths exist. Confirm the actual dataset, entitlement and integration specification for the chosen path rather than copying a workstation workflow into an API design. Both the engineering effort and the permission to use the information may differ.

07 / DecisionChoose it around a source-linked research deliverable

S&P Global is a substantial AI-related company because it combines proprietary information with technology designed for professional analysis. Capital IQ Pro is most compelling when the team needs company data and document evidence in the same research process. The product decision should follow the quality of that process under the team’s actual entitlements.

Start with one peer-comparison briefing and require another analyst to verify it. If the assistant reliably narrows the reading, preserves definitions and exposes the basis of the conclusion, the team has evidence for a wider evaluation. If it obscures those details, additional fluent output does not resolve the underlying research problem.