Key takeaways
- Atlar is now available in the Claude app directory, so any customer can connect Claude to Atlar and ask about accounts, balances, transactions, or forecasts in plain language, including receiving a spreadsheet.
- Once a workspace is connected, Claude queries live treasury data the same way a user inside Atlar would, with the same permissions, and follow-up filters pull fresh data rather than recalculating from what is already on screen.
- The connector uses Anthropic's Model Context Protocol (MCP), an open standard for connecting AI tools to external systems.
- Connecting Claude is opt-in: role-based controls apply equally to the AI, every query is logged in Atlar's audit trail, and customer data is never used to train external models.
- Atlar Intelligence is the assistant and agents inside the Atlar dashboard; the Claude integration is how the same treasury data is made available in an external AI tool the team already uses.
Atlar is now available in the Claude app directory. Any customer can connect Claude directly to Atlar and ask about their accounts, balances, transactions, or forecasts. Claude pulls the data from Atlar, returns the answer in plain language, and hands back a spreadsheet if you want one.
Many customers have been asking us for this. It's a small release with a large implication: your treasury data, complete and live, is now available from the AI tool your team is already using.
A treasury you can query from anywhere
Most AI in finance has been limited by the same thing: the data it can actually see. Exports, screenshots, and pasted figures are workarounds for the fact that the numbers live somewhere the model can't reach.
The Atlar connector for Claude removes that step. Once you've connected your workspace, Claude can query your live treasury data the same way any user inside Atlar would, with the same permissions.
Ask it to walk through your morning cash position, find every outflow over €50,000 in EUR last quarter, or pull a currency breakdown across entities. The answer comes back from the same data your treasury team is working with, with nothing to export or reformat.
It starts with a plain question. Ask how many accounts you have and where they sit, and Claude reads the answer straight from Atlar.
Ask a follow-up and Claude goes back to Atlar for the answer. Filter to one bank, switch currency, change the date range, and each response is pulled fresh from your accounts rather than recalculated from what's already on screen.
When you need to share the underlying numbers, ask Claude to put them in a spreadsheet. The file is built inside the conversation, ready to download or send on.
The same data can feed a higher-level view. Treasury at this kind of company spans several banks, currencies, and entities, and a board update usually means pulling those threads together by hand. Claude can assemble it in one pass.
All of it happens in one conversation, with the numbers as current as they are in Atlar.
Built on connectivity
The connector uses the Model Context Protocol (MCP), an open standard from Anthropic that lets AI tools connect to external systems through a common interface.
From day one, Atlar has been an API-first platform with direct connectivity to every system that touches cash. We invested in building and maintaining those connections in-house, rather than relying on third-party aggregators.
That means integrations with banks in over 100 countries, API connections to modern finance tools, and native apps for every major ERP, alongside AI-native systems like Campfire, Rillet, and Light. Together, these connections power more than €450 billion in annualized transaction volume through the Atlar platform.
In theory, any finance platform can add a connector to Claude. Whether the answer is worth acting on depends on the data behind it, and reliable, real-time data across every bank and system is what we have spent years building.
Security and controls
Connecting Claude to Atlar is opt-in, and every safeguard that applies inside Atlar applies here too. The chain of accountability extends with it.
- Claude cannot retrieve data a user doesn't already have access to. Role-based controls apply equally to the AI.
- Every query is logged in Atlar's audit trail, the same as any other access to your data.
- Customer data is never used to train external models.
For more on how we approach security and AI, read the Atlar Intelligence FAQ or visit our security hub.
Inside and outside Atlar
Atlar Intelligence is how we bring AI into the platform, with the assistant and agents built directly into the dashboard. The Claude integration is how we bring Atlar out to the AI tools your team is already using.
AI-first teams like Lovable, AMI Labs, and Parloa choose Atlar because they need connectivity they can rely on, and a platform that won't be outgrown as AI models evolve. The Claude integration is part of that.
If you'd like to see the Claude integration in action, request a demo or get in touch with our team.
