Key Takeaways:
- Panax is now available inside Claude via the Panax MCP - two offerings: Connect (global bank connectivity, standardized) and Enrich (ERP and AI-categorized, context-ready financial data).
- AI is only as useful as the data behind it. Raw bank exports give Claude something to describe. Panax Enrich gives it something to reason over.
- This is not a read-only dashboard. The MCP exposes the infrastructure layer underneath modern treasury - the same connectivity that powers the full Panax platform.
- Finance teams can now work with live cash data from inside Claude - no exports, no copy-paste, no reformatting
- The full Panax platform remains the most complete way to run treasury end-to-end; Connect and Enrich are the infrastructure, available as MCP offerings.
Treasury AI has a data problem. We fixed it.
Finance teams have been asking the same question for a while: why does AI feel so limited when it comes to actual treasury work? The answer is usually sitting in your export folder.
AI models are only as useful as what they can see. For most finance teams, what they can see is an outdated CSV pasted into a chat window, reformatted by hand. Ask Claude about your cash position and the answer comes back from a snapshot. Meanwhile, your actual accounts have moved.
There's a name for what's missing: Finance-Native AI. A model built on live, structured, categorized treasury data is a different category of tool: one where the intelligence starts at the data layer, not on top of an export. That's what we’ve built.
Panax was built to be that data layer from day one: global bank connectivity, standardized, with an AI enrichment engine that turns raw transactions into something a model can reason over.
Today, we're making it available directly in Claude.
The Panax MCP
Two offerings, available now.
Panax Connect normalizes bank data at the source. Every institution has a different format, a different API, a different set of edge cases. Connect handles that and returns a single, clean interface across accounts, currencies, and geographies. When Claude queries your balances, that's what it's reading from.
Panax Enrich takes what Connect pulls and structures it for AI. Every transaction gets categorized, labeled, and contextualized, trained specifically on treasury data. That specificity matters: generic categorization gets treasury wrong in ways that compound quietly. When Claude tells you what moved and why, it's reading data that's already been made interpretable. The model doesn't have to guess because the work is done before it asks.
Ask about your cash position across entities. Ask which outflows crossed a threshold last week. Ask for a currency breakdown before a board call. Claude goes back to your live accounts for each answer, through Panax, and returns what's actually there.
Why the infrastructure matters
Any finance platform can build a Claude connector. What it touches is the question.
Finance-Native AI requires finance-native infrastructure. Connect and Enrich are the same layer that powers the full Panax platform: direct integrations with SAP, Oracle, NetSuite, and Yardi, bank connectivity across global institutions, categorization built for treasury specifically. The MCP exposes that layer to Claude.
For teams that want everything in one place - visibility, forecasting, reconciliation, reporting - the full platform is how that works.
Connect and Enrich are available standalone for teams that want to bring Panax into the AI workflow they already have. Either way, the data is the same.
.png)
.png)
.webp)
.webp)
.png)
