ICE fixed income AI access opens through Model Context Protocol
ICE says licensed users can access its fixed income data and methodologies in AI workflows through the Model Context Protocol.
By Rafael Ortiz · Fintech Correspondent
· 3 min read
Intercontinental Exchange said its fixed income pricing data and methodologies are now available for licensed users in AI platforms through the Model Context Protocol, expanding ICE fixed income AI use cases for clients that rely on its market data. The service is currently available in Claude, with the open protocol designed to support compatibility with other AI services over time, according to ICE.
The company said the arrangement creates an AI-powered knowledge layer that allows customers to ask questions through their preferred AI environment and receive answers informed by ICE data, analytics and methodology. ICE described the move as a way to embed its fixed income information in client workflows while preserving controls over proprietary data structures and user permissions.
How will ICE fixed income data work in AI platforms?
ICE said the data is being delivered through the Model Context Protocol, or MCP, an open standard originally developed by Anthropic and governed by the Agentic AI Foundation. MCP connects large language models and AI applications to enterprise data systems, and ICE said it provides open connectivity to AI services including Claude, ChatGPT, Gemini and Microsoft Copilot.
Under the model described by ICE, the protocol includes a protected knowledge layer intended to preserve the company’s proprietary methodology, data hierarchies and relationships. Chris Edmonds, president of ICE Fixed Income and Data Services, said customers will be able to use ICE data and analytics within AI workflows, with access governed by MCP’s protections and user-permissioning features.
Edmonds said ICE is building the service around the principle that users should pay for the value of delivered data rather than the amount of computing capacity consumed. He said the approach is intended to give customers a more predictable and transparent way to add ICE data to decision-making processes.
What fixed income datasets are included?
ICE said several fixed income datasets are now available in AI environments for licensed users:
- ICE End-of-Day Fixed Income Evaluations, covering daily evaluations for more than 3 million financial instruments across more than 150 countries and more than 80 currencies.
- ICE Enhanced Evaluation Transparency, including about 160 data points per identifier across end-of-day evaluations, basic reference data, price movement commentary, mortgage-backed securities assumptive data, and trade and quote information.
- The ICE AAA Municipal Bond Curve, a transaction-driven curve available as a standalone licensed product and used in ICE’s evaluation process.
- U.S. Treasury Benchmark Data, also available as a standalone licensed product and used in ICE’s evaluation process.
- FINRA TRACE and MSRB RTRS data, covering bond trades across corporate, agency and securitized products reported to FINRA TRACE, as well as municipal bond trades reported to MSRB RTRS.
ICE also pointed to its broader work on AI tools for fixed income markets. The company said its ICE Compass platform uses AI to support buy-side fixed income trading desks with trader counterparty rankings and price estimates before trades are executed.
According to ICE, Compass combines customers’ real-time and historical trading data with ICE market data and pricing streams, as well as bids, offers, indications of interest and other pricing data received from trading counterparties. ICE said the model tracks intraday market movements, trading costs and trading behavior to produce estimates used in counterparty selection and pre-trade cost analysis.
This story draws on original reporting from Finextra Research.
