The Monetary Authority of Singapore intends to develop a know‑your‑agent protocol so banks can verify an AI agent’s identity, delegated authority, and rules. A separate sandbox will test stablecoin and tokenized‑deposit payments by AI agents as part of the Safeguards for Agentic Finance at Runtime (SAFeR) initiative. (Cryptonews.net, November 21, 2026). For finance teams managing digital assets, this is not a distant experiment; it is a clear signal that agentic finance is approaching, and that crypto treasury management must prepare for a new operational paradigm.
This article unpacks what the Singapore sandbox means for corporate and DAO treasuries, how AI agents and stablecoins will reshape treasury workflows, and the concrete steps finance teams can take today.
Table of Contents
- What Just Happened: Singapore’s AI Agent Sandbox and Stablecoin Payments
- Why It Matters for Crypto Treasury Management Now
- The Background: AI Agents Meet Crypto Treasury
- Concrete Implications for Finance Teams and DAOs
- What to Watch Next and Open Questions
- What is crypto treasury management?
- How could AI agents automate crypto treasury workflows?
- What safeguards are needed for agentic finance in treasury?
- How might stablecoin sandbox tests by central banks affect corporate adoption of crypto?
- What are the biggest risks in running a crypto treasury with AI agents?
- Key Takeaways
What Just Happened: Singapore’s AI Agent Sandbox and Stablecoin Payments
MAS managing director Chia Der Jiun outlined a two-part move: first, the creation of a know-your-agent protocol that gives financial institutions the data they need to determine an AI agent’s scope of access and assign it a secure channel; second, a sandbox that will test safeguards for AI-driven payments using stablecoins and tokenized deposits. The stablecoin payments Singapore test is designed to evaluate safeguards for AI-driven transfers. Under the protocol, a bank would receive the agent’s identity, the customer it represents, its delegated authority, and the rules it must follow—plus assurances about its origin and reliability. The customer’s mandate would specify the actions permitted and any limits, with controls scaled to the agent’s discretion and potential impact. Stablecoin Payments Now Require Agent Authority: MAS Sandbox
This announcement comes just months after BNY and Microsoft completed their first AI-to-AI treasury trade on a private ledger, as reported by CoinMarketCap. In that case, two AI agents negotiated and executed a U.S. Treasury trade without human intervention, though humans approved the final terms. Together, these events illustrate a rapid maturation: regulators are moving to build trust frameworks while institutional players already test autonomous agents in real-world treasury contexts.
Why It Matters for Crypto Treasury Management Now
For corporate crypto treasury managers, the MAS sandbox addresses a critical missing piece: a standardized way to delegate authority to software agents while maintaining compliance. Without such a protocol, businesses were exposed to unclear liability and regulatory risk if they allowed an AI agent to move stablecoins or execute cross-border payments. Now, a blueprint is emerging, and early movers can design their systems with that blueprint in mind.
Stablecoins are the natural settlement layer because they offer programmability, near-instant finality, and 24/7 operation—attributes that match the speed and rule-driven nature of AI agents. When a treasury AI moves USDC to cover a shortfall or sweeps idle balances into a yield strategy, it needs payment rails that settle immediately and can be embedded in smart contracts. Central bank sandboxes bring regulatory legitimacy, which could accelerate corporate adoption of crypto treasury automation.
The BNY-Microsoft trade shows that agentic interactions need not be limited to permissionless blockchains; however, for many companies, especially those in Web3, the combination of AI agents and stablecoin rails on public networks represents the most composable and globally accessible option for AI agents crypto treasury. The question is no longer if AI will play a role in treasury, but how to implement it safely.
The Background: AI Agents Meet Crypto Treasury

Crypto Treasury Management in 60 Seconds
Crypto treasury management is the custody, allocation, and deployment of digital assets to meet liquidity, funding, and risk objectives. It differs from traditional treasury by operating 24/7, settling on-chain, and requiring additional safeguards for key management, smart-contract risk, and regulatory compliance. Core priorities include liquidity management (balancing fiat and crypto across wallets and exchanges), risk management (counterparty, operational, regulatory), and the automation of payments and reconciliations. Crypto Treasury Management Shifts to Implementation: New EY Data shows that treasury teams are moving from exploration to active implementation, increasing the urgency for robust infrastructure.
How AI Agents Could Automate Treasury Workflows
AI agents could execute repetitive, rule-based treasury tasks with speed and consistency. Use cases include:
- Payment execution: An agent monitors accounts payable and releases stablecoin payments when invoices are approved, settling instantly across borders.
- Yield optimization: Agents sweep idle stablecoin balances into on-chain lending protocols or automated strategies, moving funds based on predefined risk parameters.
- Reconciliation: Agents match on-chain transactions with internal records in real time, flagging discrepancies.
The practical limits are real, however. As AI Agents in Treasury Explained: Use Cases, Controls, and Practical Limits notes, agents currently excel at narrow, well-defined tasks with clean data inputs. They are not ready to replace human judgment in complex strategic decisions. The MAS sandbox explicitly addresses this by requiring a customer mandate that spells out exactly what the agent may and may not do.
The BNY-Microsoft trade illustrates how agents can negotiate and settle, but the environment was controlled. In a decentralized crypto treasury, the challenges multiply: transaction fees can spike, smart contracts might be exploited, and governance processes must be integrated.
Why Stablecoins Are the Settlement Layer of Choice
Stablecoins such as USDC are programmable money. Programmable stablecoin payments allow a treasury smart contract to automatically release funds only when certain conditions are met—for example, when an AI agent’s digital signature matches the authorized list and the amount falls within a limit. They also settle in seconds, unlike bank wires that take days and are subject to banking hours. This makes them ideal for agentic workflows that operate continuously. Regulatory approvals, like the OCC’s interpretive letters allowing national banks to use stablecoins for payment activities, further pave the way. Stablecoin Payments: What the OCC’s Approval Means
Concrete Implications for Finance Teams and DAOs

The Know‑Your‑Agent Protocol and What It Demands
The proposed protocol will require treasuries to formalize their AI agent delegation. In agentic finance treasury, a bank or payment provider will demand: the agent’s identity, proof of customer authorization, a rulebook (e.g., “only move up to $50,000 USDC per day, only to pre-approved counterparty addresses”), and assurances about the agent’s software integrity. For corporates, this means integrating agent credentials into existing access control frameworks. For DAO treasury AI, it could add a new layer: the agent’s rules might be encoded on-chain and signed by a multisig, but the fiat gateway (a neo-bank or payment partner) will still need to verify via the know-your-agent protocol.
| Component | Description |
|---|---|
| Agent identity | Cryptographic proof linking the agent to its developer and version |
| Delegated authority | Customer-signed mandate detailing permitted actions, limits, and conditions |
| Rule constraints | Real-time enforcement of spending caps, approved counterparties, and time windows |
| Secure channel | Dedicated communication channel between the agent and the financial institution |
| Assurance data | Attestations about the agent’s security audit, origin, and runtime reliability |
Does Your Banking Setup Support Agentic Finance?
Assess your current financial infrastructure. Can it programmatically receive and act on instructions from an AI agent? Many corporate bank accounts lack APIs that allow third-party software to initiate payments without manual login and multi-factor authentication designed for humans. Crypto-native or neo-banking platforms that support API keys, sub-accounts, and granular role-based access control are more aligned with agentic finance. In a DAO context, where governance votes may dynamically grant an agent spending authority, the underlying banking or custody layer must reflect those permissions in real time. Why Forex Is Winning at Crypto Banking—and What You Can Learn explores how treasury teams are optimizing fiat-crypto operations, a precursor to full automation.
Three Steps to Prepare Without Over‑Engineering
- Map eligible workflows. Identify which treasury activities follow fixed rules and could be safely automated. Examples: monthly stablecoin payments to recurring vendors, daily liquidity sweeps between wallets, or FX conversions when balances cross thresholds. Not everything needs an AI agent.
- Evaluate platform readiness. Ensure your banking and custody partners offer APIs, role-based permissions, and the ability to segregate agent authority. If you use a multi-signature wallet for crypto, plan how an agent would participate—likely through a dedicated signing key that is itself governed by policy.
- Draft internal governance ahead of protocol adoption. Write a treasury policy addendum that defines how you would delegate to an AI agent: maximum transaction sizes, approved counterparty lists, approval workflows (human-in-the-loop for high-value moves), and emergency pause procedures. Use the MAS proposal as a template; you can also review lessons from past volatility in Crypto Treasury Management After the Bitcoin Crash.
What to Watch Next and Open Questions
The MAS agentic finance sandbox will run through 2027; its findings will likely shape global standards. Meanwhile, other central banks may launch similar initiatives. The commercial ecosystem will need to build know-your-agent compatible services. Several open questions remain:
- Liability: If an AI agent executes a payment that violates sanctions, who is responsible—the customer, the agent developer, or the bank?
- KYC/AML for agents: Will agents themselves need to be “identified” in a regulatory sense, or will they always be an extension of the human customer?
- Cross-border coordination: Will know-your-agent protocols be interoperable across jurisdictions, or will treasuries need to manage multiple credential sets?
As the regulatory picture evolves, keeping your organization’s knowledge base updated is challenging; tools like Joyrise: grow organic traffic while you sleep, which analyzes your site and publishes daily SEO articles, can help automate content marketing, but they are no substitute for direct monitoring of announcements from MAS and other authorities.
What is crypto treasury management?
Crypto treasury management is the practice of managing a company’s digital asset holdings—including custody, liquidity, payments, and risk—across both cryptocurrency and fiat systems to meet financial obligations and strategic goals. It extends traditional treasury by operating on 24/7 markets, settling on blockchains, and requiring specialized tools for key management and smart-contract risk.
How could AI agents automate crypto treasury workflows?
AI agents could automate rule-based tasks such as executing stablecoin payments when invoices are due, rebalancing liquidity between wallets and interest-bearing protocols, and reconciling on-chain transactions with internal ledgers. They operate according to a predefined mandate, eliminating human latency for routine actions.
What safeguards are needed for agentic finance in treasury?
Safeguards include a verifiable agent identity, cryptographic proof of delegated authority, hardcoded transaction limits and approved counterparty lists, real-time monitoring, and a human-in-the-loop mechanism for exceptions or large amounts. Treasury compliance AI can also help automate these checks, but the MAS know-your-agent protocol is designed to encode these safeguards into the banking interface itself.
How might stablecoin sandbox tests by central banks affect corporate adoption of crypto?
They reduce regulatory uncertainty, providing a clear compliance path for using stablecoins in treasury. When a central bank like the MAS validates stablecoin payments by AI agents, it signals to corporate finance teams that the practice is not a fringe experiment but a regulated activity, making it easier to get internal approval and to engage banking partners.
What are the biggest risks in running a crypto treasury with AI agents?
The biggest risks include smart contract bugs or exploits that cause fund loss, agent misconfiguration leading to unauthorized transactions, regulatory non-compliance (e.g., sanctions violations), and over-reliance on agents during market volatility when human judgment is needed. Operational risk from key compromise is also heightened when new keys are created for agent use.
Key Takeaways
- Singapore’s know-your-agent protocol and stablecoin sandbox mark the first concrete regulatory framework for AI-driven crypto treasury operations.
- Finance teams should begin mapping eligible workflows and drafting governance policies for agent delegation now, using the MAS requirements as a guide.
- Stablecoins are the natural settlement layer for agentic finance due to programmability and instant finality, but infrastructure must be upgraded to support API-driven, permissioned access.
- DAOs face unique challenges in bridging on-chain agent governance with off-chain banking verification, making crypto-friendly neo-banking platforms a critical bridge.
- The next two years will bring further central bank pilots; compliance-ready treasury designs will have a first-mover advantage.
For businesses ready to unify fiat and crypto treasury operations on a secure, API-driven platform, open a OneSafe account today.





