Hybrid [see profile for meaning] "Having personally used @CetusProtocol and others within the @SuiNetwork @SuiNetworkKR ecosystem, I went through a sequence of actions—swapping, sending, and depositing." Each service had different functions and interfaces, but the actual steps I repeated were quite similar: compare conditions to decide where to hold assets, convert them into the required tokens, review transaction details in my wallet, and approve. Once one task was done, I moved to the next service to make another decision. Going through this process repeatedly helped me better understand the Agentic Economy that Sui heavily emphasized at Basecamp 2026. This is because, if you entrust asset management to an AI agent, the more complex challenge isn’t simply executing trades faster—it’s figuring out how to translate the judgments humans currently make into clear rules for the agent to follow. For example, suppose you ask the agent: “Move my remaining funds to a reliable stablecoin yield opportunity.” The agent would need to read current interest rates and conditions, possibly exchange into another stablecoin, and then deposit it into a lending protocol. If there’s a fee to retrieve data from a service, that payment must also be included in the same process. When I do this manually, I can open each app one by one and handle them sequentially. But for automation, things change. If only the swap succeeds but the deposit fails, you end up with assets different from what you originally requested. You also need to predefine exactly how much money can be spent across multiple services. This is precisely where Sui’s Payment Intents come into play. They bundle multiple actions across different apps into a single atomic transaction. Sui’s Programmable Transaction Block can contain up to 1,024 operations within one transaction, ensuring that either all bundled actions succeed or none are applied at all. While 1,024 is a large number, having personally handled assets, I found it even more meaningful that no partial operations are left stranded upon failure. If swapping, paying, and depositing are sequential steps toward a single goal, it makes far more sense for the entire sequence to be rolled back than to leave just one step completed—this aligns better with the user’s original intent. The same applies to the scope of authority granted. If you simply hand over your wallet to an AI and let it use funds freely, it becomes difficult to trust with real assets. Sui’s concept of scoped authority imposes limits on agents—restricting how much they can spend, when they can act, and whom they can transact with—and enforces these boundaries at the protocol level. Users can specify what tasks to perform without needing to grant full, unrestricted access to their entire portfolio. As I gradually used these features, I realized there’s far more involved between the idea of “AI agents trading on their own” and the actual financial infrastructure required for Agentic Finance. In particular, a program handling money needs more than just the ability to find good trades. It must be able to complete multiple operations simultaneously, stay strictly within granted permissions, and ensure assets aren’t left in ambiguous states if something fails. Repeating countless small payments also becomes burdensome if you must prepare gas tokens for each one. This last point connects directly to features already available on Sui Mainnet. Currently on Sui, when sending supported stablecoins like USDC between wallets, you no longer need to separately maintain a SUI balance for gas fees. For an individual who sends funds occasionally, this may feel like just a convenient addition—but when a program manages multiple accounts and processes numerous small payments continuously, the entire burden of maintaining SUI balances across all wallets disappears. On-chain liquidity for trading is already in place as well. DeepBook serves as Sui’s on-chain liquidity layer, with Spot matching orders on-chain using a central limit order book model. Built on top of this same financial infrastructure are Margin (supporting up to 5x leverage) and Predict (using options structures). I found this aspect of DeepBook even more compelling than viewing it merely as a trading interface. What matters isn’t which DEX interface a person opens—it’s that apps or programs can access a shared on-chain liquidity layer whenever needed. As the primary actors in financial operations shift from humans to software, this capability becomes increasingly valuable. Hashi, which aims to connect Bitcoin, should be understood in the same context. Hashi is currently live on the Sui Testnet but is not yet available on Mainnet. When a user deposits BTC into a Hashi address on the Bitcoin network, Sui validators confirm the deposit and establish a link enabling Sui smart contracts to use that BTC as collateral.The goal is to keep actual Bitcoin on the Bitcoin network while enabling it to be used as collateral for loans and other financial services on Sui. When you examine stablecoin transfers, on-chain liquidity, Payment Intents that bundle multiple financial operations, permission management that limits agents’ spending scope, and Bitcoin-backed collateral, it becomes clear that Sui’s discussion of Agentic Finance at Basecamp isn’t merely about the word “AI.” Simply increasing transaction speed won’t suffice to delegate financial tasks currently handled manually to programs. To entrust real assets to programs, you must define the scope of funds they can access, securely complete multi-service workflows, and verify outcomes on-chain. Using Sui directly has naturally shifted my interest in this direction. I’m less curious about how many new apps are announced and more interested in understanding how much of the approval process I’ve personally gone through in my wallet is already automated in real products. At Sui Basecamp in October, I want to learn how Payment Intents are being integrated into live services, how user experience changes when agents with limited permissions move across multiple financial apps, and what Hashi is preparing to transition from Testnet to Mainnet. #Sui #SUINGAPORE