Nasdaq has launched an agentic AI environment within its Calypso capital markets platform, allowing financial institutions to connect AI agents to trading, risk and collateral data while working within controlled operational boundaries.
Announced on September 29, 2026, the launch includes a natural-language assistant that can query Calypso data and documentation. Nasdaq also plans to introduce specialized agentic workers for manual processes across capital markets and treasury operations.
The initial release is not a fully autonomous system for running the entire trade lifecycle. The capability available at launch is focused on natural-language access to information, while the broader set of operational agents remains part of Nasdaq’s planned rollout.
Agents can access Calypso data through a governed environment
The new environment is hosted through the Nasdaq Calypso cloud platform and built on Amazon Bedrock. Nasdaq says it provides sandboxing, operational boundaries and live oversight for AI agents, with no external data retention.
Agents can work with Calypso’s records covering trading, risk and collateral. The environment can also be extended with enterprise, market and reference data, according to the company.
Nasdaq is also using the Model Context Protocol (MCP) to connect external AI infrastructure to Calypso. Financial institutions will be able to integrate their own AI systems through the MCP-based layer, allowing institution-built agents to work with data and workflows exposed through the Calypso environment.
The setup is available for cloud-based Calypso deployments, while customers running Calypso on premises can connect to Nasdaq-hosted agents.
The natural-language assistant is designed for tasks such as querying Calypso information, analyzing data and reviewing documentation. Nasdaq has not disclosed the foundation model or models used by the assistant.
Broader agentic automation is still planned
Nasdaq intends to expand the environment with specialized agentic workers aimed at manual processes across capital markets and treasury. The September 29 announcement does not provide a complete list of those agents or specify which trade, collateral or settlement actions they will eventually be authorized to perform automatically.
The company’s research with Acuiti provides some context for the focus on manual work. A survey of 50 senior executives from banks, non-bank futures commission merchants, clearing brokers and buy-side firms found that 49% identified manual intervention as a significant post-trade pain point, while 56% said AI and machine learning integration was needed to avoid falling behind.
Those figures describe the broader operating environment rather than measured results from the new Calypso agents. Nasdaq has not published productivity gains, reduced settlement-failure rates or other quantified performance results attributable to the September launch.
The agentic environment follows the introduction of Calypso Intelligence in July 2026. That platform provides a governed data layer for Calypso, including standardized data models, data lineage and access controls, and was designed to support analytics and AI use cases.
Nasdaq has described the broader direction as a shift from Calypso as a “system of record” toward an “intelligent platform.” Magnus Haglind, Nasdaq’s Head of Capital Markets Technology, used that description when discussing the role of agentic AI in the platform.
The existing Calypso platform supports front-, middle- and back-office processes across areas including trading, risk, collateral, clearing and post-trade operations. The new agentic environment is being added to that existing workflow and data infrastructure rather than introduced as a separate AI product.
Nasdaq has not announced pricing or a general-availability date for the planned specialized agentic workers. It also has not said that its recent investment in AI-governance company Transient.AI is directly connected to the new Calypso environment.
Discover more from Aree Blog
Subscribe now to keep reading and get access to the full archive.
