Axelar introduced AgentFlux, an open source framework designed to run AI agents locally while keeping private keys, trading strategies, and customer data out of the cloud, a pitch aimed squarely at institutions exploring on-chain finance but wary of privacy risks.
Developed by Interop Labs, the team behind the Axelar network, AgentFlux allows financial companies to implement “agent” automation without sending sensitive information to external infrastructure, the company announced Thursday.
The framework addresses one of the biggest frictions in AI-powered crypto operations: tool calling.
Today, most agents rely on cloud-based models to decide which blockchain tools to invoke and how to structure transactions, which can unintentionally expose the very information institutions seek to protect. AgentFlux divides those tasks into two smaller specialized models: one for choosing the appropriate tool and another for generating the arguments to execute. According to the team behind Axelar, this setup improves the accuracy of tool calls by 46% in benchmarks, bringing on-premises models closer to the performance of larger cloud systems.
Sergey Gorbunov, co-founder of Axelar, shared with CoinDesk in an interview that he sees AgentFlux benefiting from multi-tasking: “First of all, there are highly sophisticated trading strategists, they are incredibly proprietary,” he said. “You definitely never want public cloud models to be as good at doing your strategy as you are. And if you upload them, you inherently like to violate them. So that’s one.”
“The second one I see is that blockchains are also used a lot for analytics. So, for example, you’re doing tax reporting or an investigation of who made what transactions. Or maybe you have a specific fingerprint and you’re expecting malicious activity in some accounts. Maybe you want to load that fingerprint into a cloud model, or AI system, and ask it to discover all the correlated transactions, all this suspicious behavior and patterns that this account has. So those types of activities you can preserve completely. And run them privately,” Gorbunov added.
AgentFlux also integrates into Axelar’s broader multi-chain strategy. The team positions Axelar as a “gateway to on-chain finance,” providing infrastructure for institutions to move assets and data across blockchains from a single integration point. AgentFlux could enable a single AI agent to see risk, assess exposure and transact across multiple ecosystems, a capability the company sees as essential for institutional adoption.
Read More: Canary Capital Files to Launch ETF Axelar Cross-Chain Tracking Protocol
