S&P Global Market Intelligence

S&P has the most rigorous retrieval design among the incumbents. Kensho's AI Data Portal answers agents with structured API calls anchored on Capital IQ identifiers, with each financial line item linked back to its source filing, and the API never sees the user's prompt. Capital IQ, Visible Alpha consensus, ratings, 12 million private companies and With Intelligence private-markets content are available in Claude, ChatGPT, Copilot, Gemini and Databricks; AI-connected clients rose about 70% in the second quarter of 2026 to over 500.

Native agents and governance documentation trail the data layer. ChatIQ Presentations is in Labs and Portfolio Insights in beta, client document ingestion in Document Intelligence has not been confirmed, and ChatIQ's model provider is undisclosed. No explicit no-training commitment, public trust centre or confirmed SOC 2 or ISO 27001 was located, and pricing is visible only through UK public-sector listings.

We initiate at 6.78, Established, with an Improving outlook. Retrieval, citations, consensus and delivery surfaces are strengths; Commercial Value (5.50) and data-governance disclosure are the weakest areas. A published no-training commitment and general availability of the Labs features would move the score toward Strong.

Strengths

• Retrieval design. Deterministic, identifier-anchored retrieval with line-item links to source filings; the API sees generated code, not prompts.
• Data breadth. Capital IQ, Visible Alpha driver-level consensus, ratings, 12m+ private companies and With Intelligence private-markets content.
• Distribution. Connectors in Claude, ChatGPT, Microsoft Copilot, Gemini Enterprise, Amazon Quick and Databricks.
• Traction. 500+ AI-connected clients and API call volume up more than five times in a quarter (company, Q2 2026).
• Evaluation heritage. S&P AI Benchmarks by Kensho (2024–25), open-sourced at retirement.
• Commitment. Kensho as a dedicated AI unit since 2018, now running the AI Data Portal.

Cautions

• Native agents. ChatIQ Presentations in Labs and Portfolio Insights in beta; agentic depth trails the data layer.
• Data governance. No explicit no-training commitment, AI subprocessor list or public trust centre located.
• Certification evidence. A UK G-Cloud listing both cites ISO 27001 and states no certification; SOC 2 not located.
• Client data. Bring-your-own-content in Document Intelligence promised for summer 2026, not confirmed.
• Model disclosure. ChatIQ's underlying model is not named.
• Organisation change. Market Intelligence reorganised in July 2026 after the Mobility spin-off.