Institutional

Nasdaq builds a sandboxed environment inside Calypso for institutions to run AI agents across the trade lifecycle

Announced on 29 September, the environment runs on Amazon Bedrock, connects agents through the Model Context Protocol and starts with a natural-language assistant. Nasdaq says more governed agents will follow.

· Source: Nasdaq


On 29 September Nasdaq announced an agentic AI operating environment inside Nasdaq Calypso, its platform for capital markets and treasury operations. The release describes a contained space in which financial institutions can run, connect and scale AI agents across the trade lifecycle. Nasdaq hosts the agents in Calypso's cloud environment, which is built on Amazon Bedrock, and agents connect to the platform through a layer based on the Model Context Protocol, which the release calls an emerging industry standard. Clients can use Nasdaq Calypso's own agents and connect their own AI infrastructure.

The first capability is modest: a natural-language assistant that lets users query the platform's data, documentation and other information. The weight of the announcement is in the controls. The release lists operational boundaries, live oversight and strict sandboxing with no external data retention, intended to keep agents inside an institution's perimeter and policies. Nasdaq cites a survey it ran with Acuiti in which 56% of respondents said they risk losing competitive ground without AI integration in clearing operations, and nearly half pointed to manual intervention as a post-trade difficulty. Magnus Haglind, Head of Capital Markets Technology at Nasdaq, says institutions need governed infrastructure inside the platforms they already run. Further governed agentic workers are planned, with no dates given.

For retail traders this is background rather than a product, but it matters for the same reason as the broker launches covered here recently: agent access is being built into the systems that sit underneath trading, with limits set by the platform rather than the user. It also extends Nasdaq's earlier move to put AI into its market surveillance platform, which this site covered in August. Two things to watch: what the later agents are allowed to do beyond answering queries, and whether the sandboxing and oversight claims are ever independently tested. The release gives no performance figures and no named client deployments for the agentic environment, so treat it as a statement of direction.


Read the original: Nasdaq - Nasdaq builds a sandboxed environment inside Calypso for institutions to run AI agents across the trade lifecycle. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.