Kensho

Paid

Kensho builds AI tools for Finance. It helps large institutions pull answers from their documents and data.

You ask questions in plain English. It searches earnings calls, filings, and reports, then returns structured answers with sources you can trace.

It only works if you already use S&P Global data products. The price starts in the mid-five figures yearly.

Open it if you run AI workflows inside a large financial institution already paying for S&P Global data. Skip it if you are a small firm, a solo analyst, or you need general-purpose AI tools.

It fits a narrow set of teams. Here is who gets value from it.

• Teams already using S&P Global CapIQ or Market Intelligence: The integration is built-in. You skip months of data mapping work.
• Firms processing hundreds of financial documents weekly: Auto-classification and red flag detection replace manual review.
• Developers building finance-specific LLM or agent applications: The LLM-Ready API gives you retrieval-augmented generation with vetted S&P data as the ground layer.
• Analysts working with filings, earnings calls, and entity intelligence: You stop reading full transcripts. You ask questions and get answers tied to speaker and timestamp.
• Investment banks and asset managers: Due diligence workflows speed up when entity linking and sentiment scoring work out of the box.
• Organizations with engineering resources for API customizations: The enterprise tier needs developers to build the last mile.

These are what keep teams using it after the setup pain.

• Natural language queries across financial documents: You type "Which pharmaceutical companies mentioned supply chain risks in their last three earnings calls?" You get a table, not a reading list. The system handles synonyms and ranks by relevance.
• Speech-to-text built for earnings calls: General transcription tools fail on financial jargon and speaker disambiguation. Scribe gets both right.
• Entity recognition that actually knows what you mean: NERD links "Apple" to the security, not the fruit, and maps it to a persistent ID in S&P's knowledge graph.
• Sentiment analysis across calls and filings: You spot trend shifts fast. Watch for sarcasm and hedged language—it still trips the scoring.
• Traceable, vetted data for AI grounding: Your LLM outputs cite S&P Global sources. Auditors and risk teams sleep better.

These limits are deal-breakers for most teams.

• No self-serve or trial: You cannot test it. You talk to sales, negotiate annually, and pay mid-five figures minimum.
• Requires engineering resources: The API is powerful but not plug-and-play. You need developers to build workflows.
• Full value needs existing S&P Global data subscriptions: Without Capital IQ, the entity linking and knowledge graph features lose most of their power.
• Specialized for financial content only: It struggles with informal or unstructured sources outside standard financial documents.
• Steep learning curve: Training and onboarding are mandatory. This is not a tool you open and use in an hour.