AI Agents Will Pay Attention

Author: Matt Levine URL: https://www.bloomberg.com/opinion/articles/2026-09-30/ai-agents-will-pay-attention One-line: AI agents will erode the profits of institutions that depend on consumer inattention (banks, credit card companies) while benefiting those that profit from action (brokerages), with Robinhood’s “Loops” agentic trading product illustrating the second model.

Key claims

• Banks are in the “inattention business” and agents threaten it. Core bank deposit profitability depends on customers leaving cash in low-yield accounts. An AI agent that optimizes a customer’s finances will move that cash to higher-yielding alternatives, compressing the spread income that makes retail banking profitable. Apollo has speculated that AI agents could trigger bank runs as consumers optimize simultaneously.
• Brokerages are in the “action business” and agents help it. More trades = more commission income for brokerages. An AI agent managing a customer’s investments will execute more frequently than a human would — the active/passive calculus shifts when execution friction disappears. Robinhood’s agentic trading product, “Loops,” is the clearest articulation of this logic: configure an objective, and the agent runs the trading strategy autonomously.
• Morgan Stanley estimates AI agents could push mortgage refinancing frequency 2× higher. The mechanism: refi is currently rate-gated and friction-gated; agents remove friction entirely. When refinancing requires only approving a workflow the agent pre-staged, consumers will refi at the optimal rate moment rather than the memorable-frustration moment.
• The same attention-destruction that harms banks harms credit card reward programs. Rewards are funded by interchange fees, which are funded by merchants who price them in. The system works because cardholders don’t maximize across all cards at all times. An agent that always routes to the best card for each transaction erodes the subsidy model (cross-citing the Yoni Rechtman “paying the agency tax” thesis).
• Manchester City was found guilty of 115 charges related to disguising £830M in shareholder funding from Abu Dhabi United Group as commercial sponsorship payments across 9 seasons, violating Premier League financial fair play rules. The club paid from ADUG directly to Man City while booking the receipts as sponsorship income from nominally independent entities.
• The CFTC dropped an enforcement case against cotton traders who had delayed reporting export contracts, raising questions about whether the regulatory retreat in financial markets extends to commodity oversight.

Notable quotes

“The basic story is: AI agents will pay attention for you. They will constantly optimize your financial decisions. This is bad for institutions whose profits depend on your inattention.”

“A bank’s deposit franchise is valuable partly because depositors leave money sitting in checking accounts earning nothing when they could move it to money market funds. An AI agent managing your finances won’t do that.”

“Robinhood announced an ‘agentic trading’ product called Loops. You configure your investing strategy and objectives, and Loops will execute trades on your behalf to pursue those objectives, checking in with you periodically.”

“Morgan Stanley thinks that AI agents could push refinancing frequency 2× higher — not because interest rates will be lower, but because friction will disappear.”

“The city paid money to Man City through intermediaries that Man City controlled, and both sides agreed to call it sponsorship income. That is not how sponsorship works.”

“The question is whether AI agents erode the spread everywhere — credit card interchange, mortgage servicing float, deposit spread — or whether they merely route the money to whoever pays agents to route it.”

How it connects

• AI Agents — the core technology whose attention-management capability is the subject of the piece
• Robinhood — building the “Loops” agentic trading product that exemplifies the brokerage opportunity
• Morgan Stanley — source of the 2× refi frequency estimate
• Apollo — speculated about AI-agent-triggered bank runs
• Financial Services — the industry most directly disrupted by attention-optimization agents
• Manchester City — the financial fair play violations case covered in the newsletter’s second story