Conversational banking AI goes off-the-shelf as MX packages it for smaller banks

Conversational assistants that answer questions about your money are not new. Cleo has done it for years, Bank of America's Erica has fielded billions of queries, and the UK neobanks from Starling to Monzo have built in-app AI over customer transaction data. What has been missing is an option for the thousands of banks and credit unions that were never going to build one themselves.

That is the gap MX is aiming at. On 30 June the financial data platform announced a conversational financial AI assistant, a white-labelled, enterprise-grade layer that institutions can drop into their own digital banking apps rather than engineer from scratch. It answers natural-language questions about a customer's finances, produces forecasts, and surfaces relevant products at the point a customer might need them.

The selling point is less the chat than the scaffolding around it. Conversations are validated against pre-configured policy rules, insights are drawn only from real-time ledger data, compliance teams keep full administrative visibility, and no outside party trains on or retains the data. Crucially, MX keeps a hard line between talking and doing: the assistant surfaces insights, but core actions like moving money or opening an account stay with the bank's existing secure processes. The system is model-agnostic, so MX can change the underlying model without changing what customers see.

"Financial institutions are sitting on incredibly powerful data. They just haven't had the right tools to act on it at the moment that matters," said MX founder and CEO Ryan Caldwell. "This assistant will change that."

For now it is an announcement rather than a deployment. MX is inviting institutions to request early access, with no launch customers named yet.

This article was prepared with the help of EDDIE, the AI editor-in-chief at Conversational AI News, and MARVIN, our AI research assistant.