Jack Paul 
Max Bareiss, Head of Lending at Galaxy Digital, thinks the next big trading client may not be a person. In comments on November 9, 2026, he argued that decentralized finance has the right plumbing for autonomous software agents that borrow, trade, and repay onchain. The point is not that humans disappear. It is that fewer steps need a human to say yes.
Why DeFi lending fits software agents
Agentic trading means software makes trading decisions and acts on them. Bareiss described a set of DeFi functions that already line up with that model. An agent can borrow assets, execute trades, open short positions, and handle repayments. Liquidation thresholds are set ahead of time and written into the protocol. Onchain liquidity is available to anyone who meets the collateral rules.
Traditional prime brokerage works differently. A person usually approves terms, monitors margin, and makes calls. For an AI agent, that person is a delay. Maybe a necessary delay in some cases, but still a delay. DeFi removes the business-hours problem.
Galaxy’s own use of onchain credit
This is not only a thought experiment for Galaxy. The firm has used Aave since at least September 2025. It borrows stablecoins against collateral such as Bitcoin and Ethereum. The DeFi work now touches treasury, trading, and structured products. The appeal is access to liquidity at any hour and less reliance on traditional intermediaries.
Galaxy has also been busy on other fronts. It announced a $100 million allocation into the Sky project around November 6, 2026. A day earlier, on November 5, 2026, the firm highlighted its Galaxy Onchain Financing Rate, or GOFR. GOFR aggregates borrowing rates across platforms including Aave, Morpho, and Spark. Galaxy also offers institutional clients first-loss capital protection and execution services in onchain money markets.
What investors should watch
For investors, the immediate question is where automated trading activity might gather. DeFi offers programmable leverage and atomic settlement. A trade and its settlement happen together or not at all. That can reduce some counterparty friction, though it does not remove market risk.
The risks need attention too. Agents that borrow and trade without human oversight depend on how well their parameters are set. Pre-set liquidation thresholds help when they work. They can also speed up losses if many automated positions hit those levels at once.
From here, watch whether GOFR becomes a reference rate across Aave, Morpho, and Spark. Watch how Galaxy deploys the Sky allocation. And watch whether other institutional lenders start treating AI agents as clients rather than tools. I think that shift will take time, perhaps longer than the current excitement suggests.