Singapore explores AI agent protocol and tokenisation framework
MAS is setting up a sandbox to test AI agents and programmable money for corporate treasury and payments.
The Monetary Authority of Singapore (MAS) will further explore artificial intelligence (AI) agents and tokenisation, setting out plans for a new agent verification protocol, a new sandbox, and updated tokenisation guides.
Whilst adoption of AI agents are still in the early stages, it has the potential to scale rapidly as the industry realizes that it can cut costs and ramp up speed of search and comparison, optimisation and decisions, negotiations and execution, said Chia Der Jiun, managing director of MAS, said in his keynote speech for Day 2 of the Singapore Fintech Festival (SFF) on 19 November.
Chia identified AI agents and tokenisation as the two transformative forces in finance.
“We do not know how quickly agentic transactions will grow, but it is likely that the future will be a mix of the human economy and a new agent economy,” Chia said.
MAS said that it will work to develop a common protocol to verify agents’ representation and mandate, as well as a sandbox to explore AI agents and programmable money, Chia said.
“The two initiatives I've highlighted, common protocols for agent transactions and the sandbox for agentic money, will be important building blocks for the future of agentic transactions,” Chia said.
The common Know-Your-Agent protocol will require customers and financial institutions to exchange information about the agent’s identity, representation, authority, mandate it operates under, and provenance.
MAS believes that this will give financial institutions the information needed to determine the terms and the scope of permission that the agents hold, Chia said.
MAS is also setting up a sandbox with the Future of Finance Institute, bringing together AI agents and programmable money for corporate treasury management and payments services, he said.
Chia recognized that AI can sharply reduce information asymmetries, reduce search and information costs in economic decisions, and enable more dynamic and continuous pricing.
For businesses, agents can automate procurement, inventory management, and other routine commercial decisions, he said.
Tokenised collateral and digital money
Chia said that MAS will work with the industry to onboard and tokenise other commonly used collateral, including foreign government securities and money market funds.
Chia announced the launch of the Guardian Tokenised Collateral Framework, which sets out the operational requirements across the collateral life cycle, including eligibility, custody, valuation, return, and enforcement.
“We are updating the guide on tokenisation of capital markets products to further clarify the regulatory and risk management requirements,” Chia said, adding that they are also working with local legal professionals to clarify issues relating to the tokenization of capital markets products.
These efforts build on MAS’ earlier issue of $100m in tokenised bills, announced by Chia during the 2025 edition of SFF.
https://sbr.com.sg/news/singapore-trial-tokenised-mas-bills-using-central-bank-digital-currency
Chia also outlined various initiatives and new frameworks in Singapore relating to tokenisation, stablecoins, and central bank digital currencies (CBDCs), including the rebranding of SGD Test Net to the SG Ledger, a settlement platform for tokenised assets, forms of digital money, and ledgers.
MAS also released the Bloom Field Guide on 19 November, which details practical considerations for deploying digital money in various cases.
Chia said that MAS tested mechanisms to settle and verify settlement conditions across separate central bank systems with the Bank of England, the Bank of Thailand, and Deutsche Bundesbank.
“This is our approach: collaborate with the industry to lay the foundations and prepare the building blocks, so that the industry can scale where there are commercially compelling use cases,” Chia said.
Chia sees the future of finance as an interoperable ecosystem involving AI agents, financial institutions, and digital money.
“Businesses, financial institutions, financial infrastructure operators will have to adapt to serving both human customers and agents,” he said.
Chia said that financial institutions will have to strengthen their value proposition, and potentially consider whether limits are needed on agent-instructed transactions to manage risk.
Regulators will have to consider the implication for financial stability as agentic use in financial markets and money movements expand.
“Importantly, consumers and businesses need to retain meaningful control over the objectives, mandate, and actions of their agents. Autonomy has to be bounded,” Chia said.