What happened
Bank of New York Mellon and Microsoft tested a setup where two AI agents negotiated a U.S. Treasury trade with each other, then humans signed off. The firms pitched it as a "Hello, world!" proof of concept, showing how AI can edge into the human-negotiated part of the market, not just algorithmic execution.
BNY calls the framework the Agentic Negotiation Protocol. Microsoft's treasury team, which holds Treasuries to manage large cash reserves, gave its agent boundaries that defined an acceptable deal. BNY added a control layer so only authorized agents could negotiate, with an overlay requiring both sides to confirm they share the same understanding of the terms before accepting them.
A transcript reviewed by Bloomberg showed the bots conducted their back-and-forth with a level of politeness that is rare on Wall Street desks.
How the firms characterize it
The companies say this lives between high-touch trading - human brokers haggling terms - and fully electronic execution. Trading desks already use AI to sift through messy, unstructured data so traders can adjust pricing, but letting computers weigh options and seek middle ground is the new step.
"We are not suggesting high-touch goes away," said Laide Majiyagbe, BNY's global head of markets and a board member at the Depository Trust & Clearing Corp. "There is that space between high-touch and electronic trading that this can step into." She added, "We believe this is the start of a really important evolution in the market."
New tools can help investors find subtle advantages in an always changing market. We're hosting our first ever investor livestream on 11/17 to show you how to use ChatGPT and Claude to invest like a pro, hosted by Briefs Finance CEO Jaspreet Singh. Save your seat.
Where BNY plans to take it and what it could mean for your portfolio
BNY plans to continue testing the Agentic Negotiation Protocol and to widen its use to negotiated workflows in repo, foreign exchange payments, collateral provisions, and contract negotiations. Majiyagbe said this could lower the "barrier to entry where human capital investment didn't make sense for certain people."
For now, it is a demo. The real challenge is proving it can scale across asset classes and function in both calm and volatile markets. A related effort: "Tradeweb Is Rolling Out AI Assistant for US Treasury Traders."
The bottom line: if agent-to-agent bargaining scales, it could bring more participants into certain corners of the market.
Curious investors often gain confidence by learning how emerging approaches can improve decisions. On 11/17, our CEO Jaspreet Singh goes live for our first ever investor livestream on how to use ChatGPT and Claude to invest like a pro. Register to join him.
