Every treasurer knows the routine. Spending hours every week downloading reports, reconciling holdings, and updating the same spreadsheets. A simple question about your portfolio can take fifteen minutes to answer.
TreasurySpring's AI-powered portfolio access changes that. You can now query your live portfolio data in natural language and get a structured answer in seconds.
Here is what it is, how it works, and three levels of advancement you can work through at your own pace.
The TreasurySpring MCP server is a secure, read-only integration that connects your portfolio data to AI assistants like Claude. MCP stands for Model Context Protocol, an open standard that lets AI tools query external systems in a structured, secure way.
In practical terms, it means you can ask questions about your portfolio in plain English and get instant, structured answers. No navigating dashboards, no applying filters, no exporting to Excel.
The integration sits alongside of TreasurySpring's existing API. It doesn't replace the API. It translates it into a form any compatible AI assistant can understand. Think of it as a universal adapter. Your systems (ERP, accounting software, TreasurySpring) all have slightly different plugs. This adapter connects them cleanly to a single AI assistant.
Through the TreasurySpring MCP server, an AI assistant can access:
A few things worth knowing upfront:
The objective is simple: replace manual lookups with natural language. Instead of downloading an Excel report and filtering it by hand, you ask a question and get an answer in seconds. Questions like:
"What have we subscribed to over the past two months?"
"Which FTFs in our portfolio mature this week?"
"Which available FTFs are compliant with our treasury policy?"
"Show our income by currency this quarter."
Not sure where to start? Connect the MCP server and ask the AI: "What questions could I ask about my TreasurySpring data?". It will tell you.
You do not need to transform your workflow overnight. We see adoption progressing through three levels, and most teams today are at level one.
Level 1: Ask questions in natural language
The starting point. Connect the MCP server, open a chat, and query your data directly. Instant answers on holdings, maturities, available cash investments and their yields, and historical positions. Setup takes less than five minutes and you could try this on your portfolio today.
Level 2: Automate recurring tasks
The next step is scheduling the tasks you do every week. With tools like Claude's co-work mode, the AI runs reports on a timetable you set:
A practical tip here: use skills files. A skill is a standardised document that teaches the AI how your team does things, including report formats, data sources, and internal conventions. Build the skill once and anyone on your team gets a consistent output every time, without re-explaining the brief. The AI can even draft and refine the skill for you.
Level 3: Multi-system automation
The most advanced level chains your systems together. Because MCP is an open standard, the same AI assistant can connect to TreasurySpring alongside your ERP, accounting software, or bank feeds, then combine them:
The principle at this level: you stay in control. The AI gathers context and prepares the work; a human reviews and approves anything that matters. As write capabilities are introduced on the TreasurySpring platform, that will mean the AI preparing a shortlist or a pre-filled subscription for you to approve. Never acting alone.
These three levels aren't a theoretical roadmap. They reflect how we see AI in institutional finance evolving: from retrieval, to automation, to orchestration across systems. We've built the integration with all three in mind so that the infrastructure is already in place as your team is ready to move through them.
AI assistants are improving quickly, but they can still make mistakes, particularly in long sessions where the model's context window, effectively its working memory, fills up. Treat the AI's output as a fast first draft and validate figures against the TreasurySpring Portal or reports, which remains your source of truth. Skills files and well-scoped questions significantly reduce errors, and the models improve with every release.
Setup takes minutes:
A step-by-step getting started guide is available, and your IT team can review the connector before rollout. View it here.
What is the TreasurySpring MCP server?
A secure, read-only integration built on the Model Context Protocol that connects AI assistants like Claude to your TreasurySpring data. It enables natural language queries of your entities, holdings, historical positions, and available fixed-term funds.
Is the TreasurySpring MCP server safe to use?
Yes. The integration is read-only, runs on TreasurySpring's authenticated API, inherits your existing user permissions, and logs every query in a full audit trail. Your data is not used to train AI models. Depending on the AI agent you use, you can also control exactly which tools the AI has access to. For example, you could grant permission to retrieve entity data only, without enabling access to holdings or available FTFs. You decide what the AI can and can't see.
Can the AI place subscriptions on my behalf?
Not currently. The connection is read-only. Write capabilities are on the roadmap and will always require human approval.
Which AI assistants work with it?
Any AI agent that supports the Model Context Protocol. We test extensively with Claude.
Do I need technical skills to use it?
No. If your IT team adds the connector once, anyone can query the portfolio in plain English.
The institutions that build AI into their treasury infrastructure now won't just be more efficient today. They'll be ready for a generation of AI tools that can prepare, recommend, and coordinate across systems, with a human always in the loop. That's the future we're building for. The read-only integration available today is the starting point, not the end state.
*TreasurySpring is the global cash investment platform transforming how institutions manage excess liquidity. It delivers simple, secure, and scalable infrastructure that enables corporations, fund managers, and other leading institutions to access 1100+ cash products through a single onboarding to its full service digital platform. TreasurySpring has facilitated $390bn+ in issuance and serves more than 900 clients across 28 jurisdictions in the UK, US, and Europe.