At Token2049 in Singapore, Hyperliquid Labs CEO Jeff Yan said the market may be making the convergence of AI and finance more complicated than it needs to be, according to on-site reporting by ChainCatcher.
Yan said agents are already superhuman in many capabilities, and he expects that gap to keep widening. From a technical standpoint, he added, there is no real barrier to agents connecting with the APIs of existing systems.
The main obstacle is not technical access
In Yan’s view, the factor that could hold back broad agent adoption is not capability or interface access. The problem is that current systems still do not place data and value on the same abstraction layer.
He said the internet can freely transmit requests and responses, but when money starts moving, the process still depends on specific institutions that have to “approve or reject” it.
Compliance may stay, but above the protocol layer
Yan said those compliance steps are likely to remain in the future. But he expects them to sit above the internet layer and the protocol layer, rather than inside the protocol layer itself.
Once the protocol layer can natively carry value the way it carries data, and compliance is handled at a higher layer, agents would be able to operate much more fully in on-chain financial systems, he said.
A future model built around user goals
Yan also said a common future setup could be one in which users define high-level goals. The purpose of financial systems, he said, is to help people build wealth and live better lives. Superintelligent systems would then turn those goals into concrete actions, while agents would play a complementary role.
That model, he added, requires the underlying system itself to be open and accessible.