Ripple on September 10 expanded its GSmart AI agent suite across the full treasury workflow of its $1 billion GTreasury acquisition, now rebranded as Ripple Treasury. The platform processes $12.5 trillion in annual payment volume for approximately 13,000 connected banks and corporate clients. New...
"Rather than asking customers to blindly trust an AI system, GSmart works within each organization's own treasury policies to surface recommendations transparently, while ensuring humans remain in control of every decision." — Renaat Ver Eecke, Senior Vice President, Ripple Treasury
Ripple on September 10 expanded its GSmart AI agent suite across the full treasury workflow of its $1 billion GTreasury acquisition, now rebranded as Ripple Treasury. The platform processes $12.5 trillion in annual payment volume for approximately 13,000 connected banks and corporate clients. New features include Knowledge Studio, a policy-interpretation engine, and Ask GSmart, a natural-language analytics layer — both requiring human approval before executing any transaction.
The release arrives as AI agent deployment in corporate treasury accelerates. In a 2026 Association for Financial Professionals survey of 240 treasury practitioners piloting AI agents, 58% reported time savings of 40% or more on cash positioning and forecast preparation. Separately, the U.S. Treasury on March 1, 2026 published the Financial Services AI Risk Management Framework (FS AI RMF), a 230-control-objective governance standard adapted from NIST's AI RMF. The framework is voluntary but expected to shape auditor standards as agent adoption scales.
Ripple Treasury is the first treasury management system with native digital asset capabilities, settling cross-border payments in three to five seconds via Ripple's RLUSD stablecoin — compared to the multi-day settlement cycles standard in traditional correspondent banking. Yet the core value proposition of this week's release is governance: deterministic financial calculations paired with policy-governed AI recommendations, positioned against a market where fewer than one in ten large global companies have deployed AI in treasury departments, according to J.P. Morgan.
The September 10 release adds orchestrated AI agents across six treasury functions: forecasting, liquidity management, risk assessment, reconciliation, financial planning, and corporate reporting. Two new components anchor the update:
Knowledge Studio is a policy module that maps every AI recommendation to a specific clause within a company's applicable treasury policy. When an agent suggests an action — for instance, rebalancing a liquidity pool or hedging an FX exposure — it cites the governing policy, allowing employees to verify the basis for the recommendation before approving or rejecting it.
Ask GSmart is a natural-language analytics assistant that accepts freeform queries against corporate financial data. It generates recommendations but routes them through the same human-approval pipeline. No recommendation executes autonomously.
The architecture separates AI from arithmetic. Machine learning handles pattern recognition, anomaly detection, and natural-language processing. Deterministic software handles all financial calculations — interest accruals, cash position aggregation, netting — to eliminate hallucination risk on numerical outputs. This bifurcation is a direct response to the well-documented problem of large language models producing plausible but incorrect numerical results.
GSmart debuted in a limited form in June 2025. This release extends agent coverage to the full treasury workflow for the first time.
Ripple disclosed two adoption benchmarks for GSmart's existing features:
The platform connects to approximately 13,000 banks and processes $12.5 trillion in annual payment volume, according to Ripple. These figures derive from GTreasury's pre-acquisition client base, which served Fortune 500 companies including American Airlines across its 40-year operating history dating to 1986.
Ripple holds 85 global regulatory licenses, a figure the company cites as relevant to the cross-border settlement capabilities it layers atop the treasury management system.
The GSmart expansion is the latest output of a three-acquisition strategy Ripple executed between 2025 and early 2026, totaling approximately $3.25 billion in disclosed deal value:
| Acquisition | Price | Closed | Function | |---|---|---|---| | Hidden Road | $1.25B | October 2025 | Prime brokerage, clearing, financing across FX, digital assets, derivatives, swaps, fixed income | | GTreasury | $1.0B | October 2025 | Enterprise treasury management (cash, liquidity, payments, netting, risk) | | Rail | Undisclosed | 2025 | Payment infrastructure |
Hidden Road, now rebranded as Ripple Prime, uses RLUSD as collateral across its prime brokerage products — the first stablecoin to enable cross-margining between digital asset and traditional markets. GTreasury, now Ripple Treasury, integrates native digital asset management alongside fiat operations.
The combined entity positions Ripple as a vertically integrated provider spanning cross-border payments (XRP/RLUSD), prime brokerage (Hidden Road), and enterprise treasury management (GTreasury) — a configuration no other crypto-native company has assembled. The economic logic: capture treasury workflow at the point where payment execution, liquidity management, and digital asset custody intersect.
The broader AI agents market reached an estimated $8.29 billion in 2025 and is projected at $12.06 billion for 2026, representing 45.5% year-over-year growth, according to Grand View Research. AI agents in financial services specifically are expected to reach $0.92 billion in North America in 2026, per Precedence Research. The overall AI agents market is projected to reach $53.2 billion by 2030 at a 44.9% CAGR.
Gartner projects the average Fortune 500 company could deploy upwards of 150,000 AI agents by 2028. Yet only 13% of companies believe they currently have adequate AI agent governance, per Gartner's same research.
Deloitte's Q4 2025 CFO Signals Survey found that 87% of CFOs expect AI to be extremely or very important to their finance department's operations in 2026. The gap between executive intent and actual deployment is significant: J.P. Morgan estimates fewer than one in ten large global companies have deployed AI in treasury departments as of mid-2026.
The addressable market is large. The corporate treasury market manages an estimated $120 trillion in global financial flows annually, according to industry estimates. AI-enabled treasury tools address a small but growing share of this flow.
The U.S. Department of the Treasury released the Financial Services AI Risk Management Framework on March 1, 2026. The document adapts NIST's AI RMF into a sector-specific standard with 230 control objectives mapped across the AI lifecycle — from data sourcing through model deployment and monitoring.
The framework is voluntary. However, Grant Thornton's analysis notes it is expected to shape auditor standards as AI adoption accelerates. Key provisions include:
Ripple's Knowledge Studio architecture aligns with the FS AI RMF's policy-mapping requirements. Each GSmart recommendation cites the governing policy clause, creating a verifiable audit trail. Whether this alignment is sufficient for enterprise compliance will depend on how auditors interpret the voluntary framework's 230 controls.
J.P. Morgan's Zack Anderson, Chief Data & Analytics Officer for Payments and Global Banking, described a scenario in a June 2026 analysis where an agentic workflow prevented a $2.3 million FX mismatch over a quarterly period through automated Monte Carlo scenario analysis. In one case study, an AI agent completed a hedge approval decision in 9 seconds — a process that typically takes hours in manual workflows.
Yet risks remain. Most corporates running agents in treasury operations have not redesigned their fraud and payment controls around AI actors, according to TrustSphere AI analysis. Existing policies and oversight structures were written for human-only operating models. The accountability question is unresolved: a treasurer can be accountable for a system they cannot fully understand, and the FS AI RMF's distinction between accountability and control does not fully address this tension.
Ripple Treasury is not operating in a vacuum. Several major players are deploying AI-powered treasury solutions:
FIS Neural Treasury launched its AI-powered suite incorporating machine learning, robotics, and the proprietary FIS Treasury GPT large language model. The platform targets automated reconciliation, real-time risk monitoring, and predictive analytics.
J.P. Morgan is developing agentic AI workflows for corporate cash management, with a focus on real-time API-based bank balance ingestion, Monte Carlo scenario generation, and NLP signal extraction. Its approach emphasizes federated learning infrastructure to address data privacy concerns.
Kyriba published an agentic finance framework for treasury teams, positioning its cloud treasury platform as an integration layer for autonomous AI workflows.
BNY Mellon deployed a settlement failure prevention tool built on Google Cloud infrastructure, targeting a different but adjacent piece of the treasury workflow.
Ripple's differentiation is the native digital asset layer — RLUSD settlement, XRP liquidity, and cross-margining through Hidden Road. No competitor currently offers treasury management, prime brokerage, and blockchain settlement in a single platform. Whether enterprises value this integration over best-of-breed alternatives from traditional financial infrastructure providers is the open question.
The GSmart expansion is functionally a governance layer wrapped around AI agents. Ripple's bet is that the primary barrier to enterprise AI adoption in treasury is not capability but trust — specifically, the ability to map every AI recommendation to a verifiable corporate policy and maintain human approval over execution. The 230-control FS AI RMF published by the U.S. Treasury in March reinforces this thesis.
The economic question is whether Ripple's vertically integrated model — combining treasury management, prime brokerage, and digital asset settlement — creates sufficient value to win enterprise clients away from established providers like FIS, Kyriba, and J.P. Morgan. The 60% and 44% adoption rates for GSmart's existing features suggest traction, but these figures measure feature activation within an inherited client base, not net-new enterprise wins attributable to the AI and digital asset capabilities.
The $12.5 trillion in annual payment volume flowing through the platform represents meaningful infrastructure. Whether AI agents can extract incremental value from that flow — through faster hedge decisions, improved cash forecasting, and reduced settlement friction — will determine whether Ripple's $3.25 billion in acquisitions generates a return proportionate to the capital deployed.