AI in treasury: a faster way to query your portfolio data

Through TreasurySpring’s Model Context Protocol (MCP) connection, clients can query their current portfolio, historical holdings, and available cash investments from compatible AI tools such as Claude.

No new treasury interface to learn. Just a new way to access and work with the information already available to you.

Ask questions such as:

• “What’s maturing in the next 30 days?”
• “Summarise my current allocation by counterparty.”
• “Show me available USD Fixed-Term Funds (FTFs) that meet my investment criteria.”

‍

What does connecting AI to treasury data actually mean?

We are now using AI tools to answer questions, analyse information, and structure data. To become useful within a treasury workflow, however, they need access to the right context.

Model Context Protocol, or MCP, is an open standard that allows AI applications to connect to external systems, and data sources.

TreasurySpring’s MCP connection provides a bridge between compatible AI assistants and the TreasurySpring platform. Once connected, a user can ask questions in natural language and retrieve information they are already authorised to access.

That currently includes:

• Current portfolio holdings
• Historical holdings
• Entities and accounts
• Available Fixed-Term Funds (FTFs)
• Data that can support reporting and portfolio analysis

For a deeper explanation of the technology and how it can be used, read our guide to connecting your TreasurySpring data to AI.

‍

What can treasury teams ask AI about their portfolio?

One of the simplest applications is replacing a series of manual lookups with a question.

Instead of finding the right report, downloading data, filtering it, and reformatting the result, a treasury team can start with what they actually want to know.

Liquidity and maturities

• “What’s maturing in the next 30 days?”
• “What’s rolling off in the next 14 days?”
• “Summarise my liquidity position across currencies.”

Understanding the timing and availability of cash remains fundamental to treasury management.

Portfolio analysis

• “Summarise my current allocation by counterparty.”
• “What’s my current allocation by currency?”
• “Summarise my portfolio by issuer type.”

This gives treasury teams another way to interrogate portfolio exposures and understand where cash is allocated across currencies, issuers and counterparties.

Cash investment discovery

• “Show me available USD FTFs that meet my investment criteria.”
• “Show me EUR FTFs with weekly liquidity.”
• “Compare the cash investments available to me by term.”

Rather than navigating individual filters, users can describe what they are looking for in natural language and use the response as a starting point for further analysis.

Reporting

• “What have we subscribed to over the past two months?”
• “Show our income by currency this quarter.”
• “Give me a breakdown of our current holdings for a portfolio report.”

The AI retrieves the relevant information through the TreasurySpring connection and presents it in response to the question.

This can compress the time between a treasury question and the information needed to answer it, while existing treasury controls and decision-making processes remain in place.

‍

How AI changes day-to-day treasury workflows

The practical impact becomes clearer when you compare the steps involved in some common treasury tasks.

Portfolio reporting

• Before:
  Download reports → filter in Excel → sense-check → reformat for internal use

• With AI-powered access:
  “Summarise my liquidity position across currencies.”

Cash investment discovery

• Before:
  Search the platform → apply filters → compare available options

• With AI-powered access:
  “Show me EUR FTFs with weekly liquidity.”

Maturity monitoring

• Before:
  Manually track upcoming maturities

• With AI-powered access:
  “What’s rolling off in the next 14 days?”

Each task may be relatively straightforward on its own. Repeated across holdings, entities, currencies, counterparties, reporting cycles, and investment decisions, however, the time spent finding and structuring information adds up. Connecting AI directly to authorised treasury data creates another route to that information: asking for it. For large corporate treasury teams, this could be particularly useful when working across multiple entities, currencies, and cash requirements.

‍

From portfolio questions to treasury workflows

Natural-language queries are the simplest place to start, but treasury teams are already exploring broader applications. We are seeing clients explore how the connection could support tasks ranging from reporting and board-pack preparation to more sophisticated workflows built around their treasury policies.

As Richard Draper, TreasurySpring’s Director of Digital Product, explains:

“Once you're connected, you can query in natural language anything about your current portfolio, historical holdings, as well as the available cash investment opportunities we have in the platform.”

TreasurySpring’s MCP connection is currently read-only. It can help users access and work with information, but it cannot place a subscription, move cash, change platform settings, or execute an investment. AI can therefore support the work around a treasury decision while the user retains control over what happens next.

‍

Is it secure to connect AI to treasury today?

Connecting AI to financial data raises legitimate questions about access, permissions, and governance. The TreasurySpring MCP connection inherits each user’s existing TreasurySpring permissions. If a user has access to particular entities or data within TreasurySpring, those permissions carry across to the connection. If they do not have access within TreasurySpring, connecting an AI assistant does not grant them additional access.

As Richard explains:

“Each user who's connecting this will have their own permissions within the TreasurySpring portal inherited.”

The connection is also read-only and built on TreasurySpring’s existing authenticated infrastructure, with an audit trail of activity. This provides a simpler way to work with treasury information while retaining the controls around who can access that information and what they can do with it.

‍

Using TreasurySpring through Claude connectors

Traditionally, adding a new capability to a treasury technology stack has often meant adding another interface: another system to log into, dashboard to navigate, or workflow for teams to learn.

Claude connectors offer a different approach. They allow Claude to access authorised information from connected applications so users can work with that information from within Claude.

TreasurySpring is available in the Claude Connector Directory, making it straightforward for clients already using Claude for finance and treasury workflows to connect their TreasurySpring data.

Users can find TreasurySpring within Claude’s connector settings, select Connect, authenticate with TreasurySpring, and begin asking questions using natural language.

Because the connection is made at user level, existing TreasurySpring permissions are inherited automatically.

MCP demo video

‍

How do I connect Claude to TreasurySpring?

For TreasurySpring clients using Claude, getting started takes just a few steps:

1. Open Settings in Claude and navigate to the connectors section.
2. Find TreasurySpring in the Claude Connector Directory.
3. Select Connect and sign in to TreasurySpring.
4. Once authenticated, start asking questions about the TreasurySpring data you have permission to access.

A useful first prompt is:

“What questions can I ask about my TreasurySpring data?”

From there, users can ask about upcoming maturities, current allocations, historical holdings, or cash investments available to them.

For technical setup instructions, see the TreasurySpring MCP documentation.

‍

What comes next for AI in treasury?

As AI in treasury develops, the interface between treasury professionals and their data is beginning to change.

Today, connecting AI to treasury data can mean asking a portfolio question and retrieving the relevant information without first finding, exporting, and manipulating the underlying data.

Over time, the same infrastructure has the potential to support more complex workflows across treasury systems, including preparing recurring reports, bringing together information from multiple sources, and preparing work for human review.

Controls remain essential. AI outputs should be reviewed and the TreasurySpring platform and its reports remain the source of truth for TreasurySpring data.

The shift is from starting with “Where do I find this?” to starting with “What do I need to know?”

‍

Frequently asked questions about AI in treasury and TreasurySpring

• Can AI place trades or move cash in TreasurySpring?

No. The TreasurySpring MCP connection is currently read-only. An AI assistant can retrieve authorised information but cannot place subscriptions, move cash, change platform settings, or execute an investment.

• Which AI tools work with TreasurySpring?

TreasurySpring’s connection uses Model Context Protocol, an open standard designed to connect AI applications with external systems and data. TreasurySpring can work with compatible AI tools that support MCP and is available directly through the Claude Connector Directory. TreasurySpring tests the connection extensively with Claude.

• Does connecting AI change my access permissions?

No. Each user’s existing TreasurySpring permissions are inherited by the connection. Users can only retrieve information they are already authorised to access within TreasurySpring.

‍

Start using AI with your TreasurySpring data

Already a TreasurySpring client?
Connect TreasurySpring through the Claude Connector Directory and start asking questions about your portfolio.

Connect TreasurySpring to Claude‍

‍

For more advanced examples, technical information, and ideas for building AI into your treasury workflows, read Connect your TreasurySpring data to AI: a guide to the TreasurySpring MCP server.