According to Axelar co-founder Sergey Gorbunov, AgentFlux enhances tool-calling accuracy and enables secure AI automation for blockchain finance without exposing proprietary strategies or sensitive data to cloud systems.
Interop Labs officially announced AgentFlux, an open-source framework enabling local deployment of AI agents while safeguarding private keys, trading strategies, and client data from cloud exposure. Developed by the team behind Axelar, AgentFlux addresses tool-calling inefficiencies by splitting tasks into specialized models, improving accuracy by 46% on benchmarks. Sergey Gorbunov highlighted its applicability for confidential trading strategies and blockchain-based analytics such as tax reporting or transaction investigations. Integrated into Axelar’s multichain network, AgentFlux allows institutions to view risk, assess exposure, and transact across ecosystems from a single point, supporting private, secure onchain finance automation.