Ripple is putting a broader layer of AI agents into the corporate treasury software business it acquired for $1 billion last year, giving investors a clearer look at what the company’s expansion beyond crypto payments is actually producing.
The company expanded GSmart, the AI system embedded in Ripple Treasury, with agents covering cash forecasting, liquidity management, account reconciliation, risk monitoring and financial reporting. The former GTreasury platform was acquired by Ripple in October 2025 in a $1 billion deal and has since been rebranded as Ripple Treasury.
GSmart itself is not entirely new. GTreasury launched the technology before Ripple bought the company. The significance of the latest release is that Ripple is now extending those capabilities across the treasury platform it paid $1 billion to enter, turning the acquisition into a broader enterprise software bet built around AI as well as digital assets.
Ripple’s $1 Billion Treasury Bet Is Taking Shape
When Ripple announced the GTreasury acquisition in October, the strategic case centered on entering corporate treasury management and connecting conventional cash management with stablecoins, tokenized assets and faster payments.
GTreasury brought more than four decades of treasury software experience and connectivity across banks and enterprise resource planning systems. Ripple’s pitch was that adding its payments and digital-asset infrastructure could eventually give finance departments one platform for managing both traditional and digital money.
The latest GSmart expansion adds another layer to that thesis.
Ripple says the agents monitor treasury processes and propose actions across forecasting, planning, liquidity, risk, reconciliation and reporting. Knowledge Studio acts as the policy layer, allowing a company to encode internal limits and controls that govern what an agent can recommend. Analytics Studio includes Ask GSmart, a conversational tool for querying treasury data and producing reports.
The AI Can Recommend, but It Cannot Approve Its Own Move
The more interesting part of Ripple’s design is where it deliberately limits AI.
Each agent can identify an issue and propose an action, but it must cite the specific company policy supporting that recommendation and wait for human approval before execution. Ripple Treasury says the proposal, policy citation, approver and timestamp are retained for the audit trail.
The financial calculations are also kept separate from generative AI.
Ripple Treasury says figures used inside proposals are produced by deterministic software rather than generated by the AI model. That distinction matters in corporate treasury, where a hallucinated explanation is inconvenient but an invented cash balance, hedge amount or payment figure can become a financial-control problem.
The structure effectively divides the job: conventional software handles the maths, AI interprets data and policy, and a human remains responsible for approving an actual financial action.
Ripple Says Customers Are Already Turning the Tools On
Ripple is also offering its first indicators of adoption, although the numbers are company-reported and apply only to customers eligible for the features.
The company says 60% of eligible customers have enabled Risk Insights, which identifies exposure anomalies and potential policy breaches, while 44% use Forecast Insights to compare projected cash flows with actual results and surface emerging liquidity gaps. Ripple has not disclosed the underlying eligible-customer counts, so those percentages should not be read as adoption across its entire customer base.
That leaves revenue, contract values and the financial return on Ripple’s $1 billion acquisition undisclosed.
But the product direction is becoming clearer. Ripple did not spend $1 billion merely to attach crypto rails to an existing treasury system. It bought an enterprise software platform and is now pushing AI deeper into the workflows where corporate cash, risk and payments are actually managed.
The notable part is that the crypto company is increasingly selling that expansion not on XRP or token prices, but on software, governance and automation.