Ripple announced on September 10 that it has expanded GSmart, the AI capability inside Ripple Treasury, with orchestrated agents for forecasting, liquidity, risk, reconciliation and reporting. The company calls it the industry’s first governed AI for enterprise treasury, though that claim has not been independently established.
GSmart itself is not new. GTreasury introduced the platform in June 2025, before Ripple acquired the company. The latest release adds a larger catalogue of task-specific agents, a policy layer called Knowledge Studio and Ask GSmart, a conversational interface for treasury data and reporting.
The change is that GSmart now connects separate AI features through a common policy and approval layer. Rather than treating forecasting, risk or reporting as isolated tools, Ripple is placing them inside the workflows treasury teams already use to review cash, liquidity and financial exposures.
How Ripple says GSmart handles a proposed treasury move
It separates the part of treasury work that needs mathematical certainty from the part where AI may be useful. Deterministic software performs the financial calculations, while GSmart interprets policies, identifies patterns and explains the proposed action. Knowledge Studio allows companies to define the controls against which those proposals are checked.
In practice, a liquidity agent could identify an imbalance between accounts, while a risk tool may flag an exposure anomaly or a policy breach. The system is designed to cite the rule behind its recommendation, leaving the treasury team to decide whether the action should go ahead.
That distinction matters because the announcement does not describe a system that freely reallocates corporate cash or trades digital assets. It describes a recommendation system built to operate within the approvals and audit requirements already used by finance teams.
Ripple’s $1 billion acquisition of GTreasury, announced in October 2025, supplied the distribution for that strategy. The deal gave Ripple a treasury platform serving more than 1,000 customers across 160 countries, together with more than four decades of experience in corporate treasury operations.
GTreasury already connects treasury and finance teams with banks and enterprise resource-planning systems, giving customers a consolidated view of data needed for payments, forecasting and risk management. Those links to existing financial processes are what make an AI layer useful: it is being inserted into an established operating environment rather than offered as a separate chatbot.
That is the strategic value beyond the software itself. If GSmart becomes part of the routine process through which customers assess cash positions and liquidity needs, Ripple moves closer to the point where financial decisions are made.
Ripple Treasury has already begun building the infrastructure beneath that decision layer. In April, the company launched capabilities that let customers view and manage bank cash, XRP, RLUSD and assets held through external custodians in one system. As Coindoo covered in its report on Ripple’s unified cash and digital-asset treasury platform, the aim is to bring onchain and traditional balances into the same treasury environment.
The company says its treasury platform facilitated $13 trillion in customer payment volume during 2025. That figure does not show how much money GSmart influences or manages, but it illustrates the scale of the existing system to which Ripple is adding AI capabilities.
Approved recommendations could eventually connect with Ripple’s payment, custody and digital-asset services where those products fit a customer’s needs and are available. That should not be read as a guarantee that every GSmart user will use stablecoins or XRP. The more immediate point is that a recommendation about liquidity, payments or idle cash can be considered alongside both traditional and digital-asset balances.
Ripple’s emphasis on controls reflects a broader problem with enterprise AI. In a survey of 360 IT application leaders, Gartner found that only 15% were considering, piloting or deploying fully autonomous AI agents. Just 19% said they had high or complete trust in vendors’ hallucination protections, while 74% viewed agents as a new attack vector.
Those findings help explain why Ripple is marketing traceable recommendations and human sign-off rather than maximum autonomy. In treasury, a poor recommendation can affect payment timing, liquidity, compliance and financial reporting, so the source of a proposed action matters as much as the action itself.
The controls are only as reliable as the customer’s underlying data and policy settings. A cited rule may make a recommendation easier to review, but it cannot compensate for inaccurate inputs or poorly configured limits.
The company says 60% of eligible customers have enabled Risk Insights and 44% are leveraging Forecast Insights. Those figures show that users are turning on parts of GSmart, but they do not reveal how many customers were eligible, how frequently the tools are used or how often recommendations result in approved action.
The company has also not provided results on time saved, forecast accuracy, error reduction, financial returns or the value of funds influenced by GSmart. Without those measures, feature enablement is not evidence that the AI system has improved treasury performance.
Ripple has built a system designed to place AI inside existing corporate treasury controls rather than outside them. Its $1 billion GTreasury acquisition gave the company the customer base and financial workflows needed to attempt that strategy; GSmart is the decision layer being added on top.
The next proof point is more concrete than another feature rollout. Ripple will need to show whether treasury teams accept the recommendations, how often those approvals lead to action and whether customers see measurable improvements in forecasting, liquidity management, risk monitoring or operational speed.
This article is for informational purposes only and does not constitute financial, investment or treasury-management advice.
The post Ripple Adds Governed AI to Its $1B Treasury Bet appeared first on Coindoo.
