AI Agents Move Into Personal Finance as Meta Tests Muse and Banks Weigh Overhaul

The consumer test followed Meta chief AI officer Alexandr Wang's promotion of the #MuseMoneyChallenge, which encouraged users to put Meta's new AI agent to work finding $1,000 in savings. Early examples shared around the challenge included cheaper insurance, forgotten subscriptions and unused gift cards, though TechRadar reported that many of the early examples highlighted came from Meta employees.

Muse is different from a conventional chatbot because Meta designed it to take action rather than simply tell users what to do, according to TechRadar. The agent runs on its own secure virtual computer with a browser, can connect to services such as email and calendar, and can handle tasks including filling out forms and dealing with customer service. It can continue working after the app is closed, although purchases and sending messages can require user approval.

The TechRadar writer gave Muse broad access to look for savings. Meta lets Muse connect to other apps and services, and says users control those permissions. The company also says login credentials are stored separately so the agent cannot read them, while sensitive actions can require approval.

Muse eventually identified some Patreon and Substack subscriptions the writer might not want anymore and an opportunity to reduce a phone bill based on newly released data plans. The useful savings came to about $35, or 3.5 percent of the $1,000 goal. The writer checked the suggestions before acting on them and said the result was mildly impressive because it was still $35 that would probably have continued to be spent each month otherwise.

The writer concluded that the gap between the personal result and the #MuseMoneyChallenge showed how much an agent's usefulness can depend on the amount of financial debris a person has accumulated. Recovering a year's worth of subscription payments is possible if someone has accidentally been paying for something for a year, but those opportunities are not necessarily waiting in everyone's accounts. The writer said a better use of Muse might be searching for refunds and credits, or handling administrative tasks that are postponed because the potential reward does not seem worth 40 minutes navigating a company's customer service system.

In the separate TechRadar article, the founder of Molit.ai and Social Discovery Ventures wrote that banking technology has for decades followed a simple sequence: a person makes a financial decision and the bank's technology processes it. AI can potentially reverse that sequence by interpreting a person's goals, understanding the financial context and determining what happens next. It could recognize that a customer is likely to face a cash shortfall, identify available ways to address it and potentially execute the appropriate action.

McKinsey estimates that generative AI could create $200 billion to $340 billion in annual value for the banking industry, according to the article. Yet much of the industry is adding AI to systems designed for the old model rather than rebuilding the model around AI, the article said. The problem is that AI inherits the same product silos and process boundaries, giving banks another layer of technology but not the cross-system decision-making needed to realize AI's full potential.

The article said the most visible AI applications in banking are often the easiest to deploy. Banks are using AI tools to improve customer service, automate fraud detection, personalize recommendations, summarize documents and accelerate credit decisions. Lloyds Banking Group said more than 50 AI use cases were rolled out across the group in 2025, generating around £50 million in value, with more than £100 million in additional value expected in 2026.

McKinsey has made a similar observation, saying that simply adding AI on top of existing processes will not produce transformational change and can instead create another layer of technical debt, according to the article. The article said the difference between an AI-native architecture and a chatbot attached to an existing system can be tested by asking whether the AI can choose and coordinate actions within guardrails, trusted data sources and approved tools; whether the process is designed for the agent to act with human review and recorded decisions; and whether permissions, monitoring and regulatory controls are built into the workflow, particularly for compliance and fraud prevention. If the answer is no, the company has added an AI interface without redesigning the underlying process.

The article said more significant transformation begins when a bank starts with the customer's objective, not with the banking product. For a customer who wants to maintain liquidity while earning as much as possible on excess cash, an intelligent banking system could continuously monitor balance, upcoming payments, income, available credit and other relevant information, then decide whether money should remain liquid, be invested elsewhere or be used to reduce borrowing. Adoption could begin with low-risk actions such as moving excess cash into savings or setting aside VAT for future tax payments before expanding into more consequential decisions.

Deutsche Bank has deployed an agentic AI system for third-party risk management in which several AI agents retrieve relevant controls, analyze supporting documentation and propose assessment outcomes, according to the article. Human assessors remain responsible for reviewing or overriding those recommendations. The article said AI agents should receive defined permissions like a new employee, with transparent activity logs, alerts, approval thresholds and the ability to override decisions.

The article also noted that Visa and Mastercard are building infrastructure for AI-initiated payments, allowing agents to act on behalf of consumers and businesses. Visa Intelligent Commerce is designed to let AI agents find and purchase products on a user's behalf, with tokenized credentials, authentication and spending controls built into the payment flow. As agents move from recommending actions to executing them, banks will need to ensure transactions remain within the customer's intent and risk tolerance.