Adaptive Retrieval
Adaptive Retrieval (formerly known as Kensho Grounding Agent) is a data retrieval service that enables accurate, trusted access to S&P Global data from any GenAI application. Given a natural-language query, the solution searches across S&P Global's AI-ready datasets, identifies the specific information needed to answer the question, and returns results with citations.
The Adaptive Retrieval solution is designed to serve as a single point of entry across multiple S&P datasets by translating user intent into structured retrieval tasks.
We are continually integrating more S&P datasets into the solution's knowledge base to make it smarter and more capable.
Key Benefits
Adaptive Retrieval serves as a retrieval layer within larger systems, not as a standalone tool. It can power agents through complex multi-agentic workflows. Key benefits include:
- Provides citations to enable auditability — Every response includes source links to S&P Global datasets
- Facilitates seamless data retrieval across S&P Global datasets — Access multiple datasets through a single API
- Integrates into any GenAI application — Works with any LLM or agentic framework
- Reliable and trustworthy outputs — Data comes directly from S&P Global's trusted databases
Example Use Cases
Chatbots or Analyst Assistants
Use Adaptive Retrieval to power grounded answers to factual, data-intensive questions, with source links users can verify.
Report Generation or Automation Agents
Integrate Adaptive Retrieval with agents that populate business reports, summaries, or dashboards ensuring the content is backed by trusted data sources.
Deep Research Agents
When conducting multi-turn or in-depth research, delegate factual lookups (e.g., KPIs, historical metrics) to Adaptive Retrieval, while other agents handle synthesis or recommendations.
Getting Started
To receive access, please email commercial@kensho.com for information on trials and pricing.