Kensho
EnterpriseS&P Global's AI and machine learning arm for financial research, entity linking, and document analytics
What is Kensho?
Kensho is the AI and analytics division of S&P Global, acquired in 2018 for about $550 million. It builds machine learning products aimed at the toughest problems in institutional finance: extracting structured data from unstructured filings, linking entities across millions of documents, transcribing earnings calls, classifying news, and generating natural language answers over financial knowledge graphs. Kensho's core products include Scribe (a finance-tuned speech-to-text engine trained on thousands of hours of earnings calls and accented financial vocabulary), NERD (a named entity recognition service that maps mentions of companies, people, and securities to S&P Global's permanent IDs), Classify (automated tagging of documents by industry, topic, and event type), Link (entity linking across the S&P Global knowledge graph), and LLM-Ready API (which turns S&P data into context objects for retrieval-augmented generation). Kensho does not sell to retail investors — its customers are banks, hedge funds, asset managers, regulators, and the internal teams at S&P Global that power Capital IQ, Market Intelligence, and Platts. Pricing is enterprise-only and typically negotiated through an S&P Global account rep as part of a broader data subscription. If you already pay for Capital IQ or S&P data feeds, Kensho's APIs are usually the natural next step for adding machine learning to your workflow.
⚡ Quick Verdict
Banks, hedge funds, and asset managers already using S&P Global Market Intelligence or Capital IQ
Retail investors, individual traders, or anyone needing self-serve pricing
Enterprise only · Custom pricing via S&P Global
No — enterprise contracts only
Deep integration with S&P Global's data — Scribe, NERD, and Classify are battle-tested on institutional workloads
No public pricing, no free tier, and onboarding runs through enterprise sales cycles
Bottom line: Scores 4.4/5 — The gold standard for AI on S&P Global data, but only accessible to institutional clients. If you're not already an S&P customer, look at AlphaSense or Sentieo instead.
Pricing
Enterprise only: Kensho does not publish pricing. All products (Scribe, NERD, Classify, Link, LLM-Ready API) are sold through S&P Global sales reps as add-ons to existing data subscriptions. Pilots are typically 3–6 months. Multi-year contracts are the norm for hedge funds and asset managers. If you're an existing S&P Market Intelligence or Capital IQ customer, ask your rep about Kensho API access — it's often bundled into enterprise data packages.
Key Features
- Scribe — finance-tuned speech-to-text for earnings calls
- NERD — named entity recognition mapped to S&P permanent IDs
- Classify — automated document tagging by industry and topic
- Link — entity linking across the S&P Global knowledge graph
- LLM-Ready API — retrieval-augmented context from S&P data
- Integration with S&P Capital IQ Pro and Market Intelligence
- On-premise and private cloud deployment for regulated clients
- SOC 2 Type II, ISO 27001, enterprise security posture
Pros & Cons
Pros
- Models trained on real institutional finance text, not generic web corpora
- First-party integration with S&P Global data assets
- Enterprise security, compliance, and dedicated support
- Strong NLP accuracy on accented speakers and financial jargon
Cons
- Enterprise-only — no self-serve API or free tier
- Locked to the S&P Global ecosystem for best value
- Opaque pricing requires a sales cycle to evaluate
FAQ
Is Kensho available for individual investors?
No. Kensho is sold exclusively through S&P Global as an enterprise product to banks, hedge funds, asset managers, and data teams. There is no self-serve signup, no free tier, and no trial. Individual investors who want AI-powered financial research should look at AlphaSense, Sentieo, or Simply Wall St instead — all of which offer public pricing and free or low-cost entry tiers.
What is Kensho Scribe and how accurate is it?
Scribe is Kensho's automatic speech recognition model, purpose-built for finance audio like earnings calls, investor days, and macro panels. Kensho claims Scribe outperforms general-purpose ASR on financial vocabulary and accented speakers because it's trained on thousands of hours of transcribed S&P earnings content. Accuracy in the high 90s is typical for clean US English audio, dropping for noisy or heavily accented speakers.
How does Kensho compare to AlphaSense?
AlphaSense is a full market intelligence search product with self-serve pricing, a polished UI, and broad content coverage including analyst reports, filings, and expert transcripts. Kensho is primarily a set of APIs and ML components — you bring the workflow, Kensho provides the models. If you want a finished research tool, pick AlphaSense. If you want to build your own, Kensho is closer to your need.
Does Kensho offer a generative AI product?
Yes. The LLM-Ready API turns S&P Global structured and unstructured data into retrieval contexts for your own large language models. Kensho also offers benchmarks and evaluation harnesses for finance LLMs. It does not sell a standalone chatbot — the idea is that Kensho makes S&P data usable inside whatever LLM stack you choose (OpenAI, Anthropic, internal models, etc.).
Can hedge funds use Kensho for alpha generation?
Yes, and many already do. Quant funds use Scribe transcripts for NLP signal extraction, NERD for entity linking across filings, and Classify to tag events like M&A, guidance changes, or management turnover. Kensho's data lineage back to S&P Global is a selling point for compliance and auditability. Funds still need to build their own signals on top — Kensho is the plumbing, not the alpha.
Is Kensho SOC 2 and GDPR compliant?
Yes. As part of S&P Global, Kensho inherits enterprise-grade compliance including SOC 2 Type II, ISO 27001, and GDPR. Data processing can be scoped to specific regions, and on-premise or private cloud deployments are available for clients with strict data residency or regulatory requirements.
Who are Kensho's main competitors?
For institutional NLP on financial text, competitors include Bloomberg's internal NLP stack, FactSet's Truvalue Labs, RavenPack, and AlphaSense's API business. For entity linking and knowledge graphs, Refinitiv (LSEG) and FactSet Entity Resolution are common alternatives. Kensho's moat is its direct pipeline to S&P Global data and permanent IDs.
📋 Good to know
Contact an S&P Global sales rep. Kensho products are typically scoped during a 30–60 day discovery and then piloted for 3–6 months before a production rollout.
SOC 2 Type II, ISO 27001, GDPR. On-premise and private cloud available for regulated clients.
Kensho only makes sense if you already use S&P data or need its specific ML primitives.
High — Kensho products are APIs aimed at engineering and data science teams.