Be Resilient

JANUARY 2027: PROTOCOL-LEVEL LOCKS ARE COMING TO MAINNET AI agents are becoming increasingly capable of executing transactions, making purchases and interacting with financial infrastructure. But there's an important question behind agentic payments: How do you give an AI agent permission to spend money without giving it unrestricted control over your funds? At the recent @Concordium Town Hall, @PeterMarirosans , Concordium's CTO, explained how Protocol-Level Locks (PLLs) are designed to address this. The idea is straightforward: define the spending permissions at the protocol level, while retaining custody of your tokens. Here's how it works: → 1. CREATE A LOCK A user creates a lock and defines its permissions. Three important decisions can be configured: → Who can fund the lock. → Who is authorised to spend from it. → Where the funds are allowed to go. These rules are established at the protocol level, rather than relying on a conventional smart contract to hold the funds. → 2. RETAIN CUSTODY OF YOUR TOKENS This is one of the important distinctions Peter highlighted. Concordium's Protocol-Level Tokens (PLTs) remain in the user's custody. Funding a lock does not mean transferring the tokens into a smart contract or handing ownership over to an agent. The user retains control while establishing the conditions under which spending can take place. → 3. GIVE YOUR AI AGENT DEFINED SPENDING PERMISSIONS Consider a practical example. You create a shopping agent and authorise it to make purchases on your behalf. But instead of giving it unrestricted spending authority, you configure its lock: → Supermarket A: Allowed. → Supermarket B: Allowed. → Supermarket C: Allowed. → Supermarket D: Not allowed. Now, if the agent attempts to make a payment at Supermarket D, the transaction fails. Why? Because the destination does not satisfy the permissions defined in the lock. The agent can execute transactions, but it cannot independently expand the authority granted to it. → 4. EXTEND THE LOGIC WITH SMART CONTRACTS The initial functionality establishes who can fund, spend and receive funds. Peter also explained how additional programmable conditions can be introduced through smart contract integration. This could enable configurations such as: → Daily spending limits. → Weekly or monthly spending limits. → Different spending conditions. → Additional triggers and transaction configurations. This introduces another layer of control over how autonomous agents interact with financial resources. → 5. SECURITY BEFORE DEPLOYMENT Peter also addressed the security considerations behind the development. Concordium has undertaken extensive security work with @CertiK , alongside internal security reviews, to ensure the protocol-level implementation is properly assessed before release. The delay in delivery has been driven by the need to validate the security and functionality rather than rush deployment. → THE ROLLOUT December 2026 → Protocol-Level Locks arrive on testnet. 27 January 2027 → Planned mainnet launch. WHY THIS MATTERS FOR THE AGENTIC ECONOMY There is a difference between allowing an AI agent to execute a payment and defining the boundaries within which it is authorised to operate. An agent needs the ability to act. The owner needs the ability to define what that agent is allowed to do. Protocol-Level Locks introduce this distinction at the infrastructure level: PLTs → The underlying tokens. Protocol-Level Locks → Spending permissions and destination restrictions. AI Agents → Execution within the authority granted to them. Smart Contracts → Additional programmable conditions and spending logic. This is an important piece of the infrastructure required for agentic commerce. Because the future of autonomous payments shouldn't depend on simply trusting that an AI agent will behave as instructed. It should also have protocol-level mechanisms that enforce the permissions it has been given. concordium:native