The arrival of Meta’s Muse artificial intelligence agent has introduced a new source of uncertainty into financial markets: What happens to banks, insurers, brokerages and wealth management firms when consumers can delegate increasingly sophisticated financial decisions to a personal AI assistant?
The question goes beyond whether AI can automate customer service, summarize investment research or help financial advisors prepare for client meetings. It concerns the possibility that a general-purpose AI agent could become the primary interface through which consumers manage their financial lives, potentially changing how financial institutions acquire customers, distribute products, earn fees and maintain client relationships.
For financial services executives, the distinction is consequential. An AI assistant that helps an advisor serve clients more efficiently strengthens an existing business model. An AI agent that independently helps consumers compare financial products, select service providers and execute transactions could change the economics of that model.
Muse has brought these competing visions of financial AI into sharper focus. Its emergence also comes as Anthropic, OpenAI and other technology companies are developing increasingly specialized AI capabilities for financial professionals.
The resulting debate is not simply about whether machines will replace human financial advisors. It is about who will control the relationship between financial institutions and their customers in an economy increasingly organized around AI agents.
Wall Street Takes Notice
The immediate catalyst was a selloff in U.S. financial stocks during the past week, as investors reassessed the competitive position of established financial institutions following Muse’s rapid adoption.
Shares of Charles Schwab, LPL Financial and other financial advice businesses came under pressure as investors considered whether AI agents could reduce the value of services traditionally delivered through brokerages and human advisors. Banks, insurers and travel companies also faced selling pressure amid concerns that automated consumer agents could make it easier to compare prices, negotiate with providers and switch between competing services.
The market reaction reflected a broader divergence in investors’ expectations for AI. While established service providers faced questions about potential disruption, Meta attracted enthusiasm over Muse’s early adoption and its potential to create new subscription, commerce and transaction revenue.
For brokerages, one particularly sensitive issue is cash management. An AI agent with access to a consumer’s financial accounts could identify idle cash earning relatively little interest and suggest moving it into higher-yielding alternatives. If consumers followed those recommendations at scale, brokerages that benefit from customer cash balances could experience pressure on an important revenue source. That possibility helps explain the attention directed toward Schwab.
For financial advice businesses, the threat is different but related. An AI agent capable of assembling a household balance sheet, evaluating investment costs, identifying tax-planning opportunities and comparing retirement strategies could reduce the amount consumers are willing to pay for certain financial planning services. The market is not necessarily predicting the disappearance of human advisors. Rather, investors are reconsidering how much of the financial advice value chain might eventually be automated, how quickly that automation could spread and whether incumbent firms would retain their existing profit margins.
The reaction also recalls earlier financial-sector volatility associated with AI developments. Financial stocks came under pressure earlier in 2026 after wealthtech company Altruist introduced an AI-powered tax-planning capability, raising questions about the competitive position of established financial planning and brokerage providers. Muse has broadened that debate because it is not a specialized application designed exclusively for financial professionals. It is a general-purpose consumer agent developed by one of the world’s largest technology companies.
That distinction changes the potential scale of the competitive challenge. A specialized financial AI tool generally must persuade an institution or advisor to adopt it. A widely used consumer agent could instead become part of a customer’s everyday digital environment, influencing financial decisions before the customer contacts a bank, insurer or financial advisor. The financial consequences will depend on whether consumers trust AI agents with sensitive information, whether those agents can consistently deliver accurate financial guidance and whether regulators permit them to undertake increasingly consequential financial activities.
What Is Muse?
Meta introduced Muse on September 8 as a personal AI agent designed to perform tasks on behalf of consumers rather than merely answer questions.
The company describes Muse as a system capable of coordinating activities across applications, remembering users’ preferences, developing plans and working toward longer-term goals. It is available through a dedicated application and WhatsApp, with an initial rollout in the United States across iOS, Android and the web. Meta has also announced plans to bring Muse to its AI glasses.
Muse represents an evolution from conversational AI toward what the technology industry calls agentic AI.
A conventional chatbot primarily responds to a prompt by generating information, explanations or recommendations. An AI agent can take additional steps toward completing the underlying task, potentially using software tools, retrieving information, interacting with websites and executing authorized actions.
Consider a consumer who wants to reduce monthly household expenses. A conventional chatbot might recommend reviewing subscriptions, shopping for cheaper insurance and comparing credit card fees. An agent such as Muse can potentially perform parts of that work directly, identifying expenses, contacting service providers, negotiating prices and helping the consumer complete transactions.
The distinction is between providing instructions and carrying out the work.
Meta Chief AI Officer Alexandr Wang has promoted this capability through the #MuseMoneyChallenge, highlighting examples of the agent identifying unused subscriptions, negotiating lower bills and finding discounts or refunds. The company’s promotional examples illustrate the potential financial applications, although individual savings claims should not be treated as evidence of typical results.
Muse is intended for a broad consumer audience rather than a particular professional occupation. Meta offers a free version alongside subscription tiers priced at $20 and $100 per month, according to Reuters. The company is positioning the agent as an everyday assistant capable of handling activities ranging from personal productivity and travel arrangements to shopping and financial management.
Its early adoption has attracted considerable attention. Reuters reported that Muse recorded approximately 2.8 million downloads during its first 12 days, a figure that helped fuel enthusiasm for Meta’s AI strategy. Analysts at Jefferies and JPMorgan subsequently highlighted its potential to generate revenue through subscriptions and commercial transactions.
Downloads, however, are an early indicator of consumer interest rather than a reliable measure of sustained engagement, profitability or successful financial outcomes. The more consequential questions concern whether users continue delegating tasks to Muse after the initial novelty fades and whether Meta can convert that activity into recurring revenue.
How Muse works
Muse is powered by Muse Spark, an AI model developed for tasks requiring reasoning, tool use and coordination across multiple steps.
Rather than operating solely within a conversational interface, Muse uses a dedicated virtual computing environment that allows it to interact with websites and connected applications on a user’s behalf.
Meta calls this environment Muse Secure VM. It provides an isolated cloud-based computer containing the agent and the information required to perform authorized tasks. Within that environment, Muse can use a browser, complete forms, communicate with services and continue working after a user closes the application.
The system also incorporates a separate security agent, called Sentinel, that reviews actions before they reach the internet. Meta says users retain control over connected services and permissions, with additional approval required for sensitive activities such as sending emails or making purchases.
According to the company, credentials are stored separately from the agent, and users can review an audit trail of completed and planned activities. Meta also says that conversations and information stored within Muse’s virtual environment are not shared with its advertising systems.
These protections are particularly relevant to financial applications. An AI agent that can access banking information, interact with financial websites and initiate transactions presents different security challenges from a chatbot that merely explains the difference between a stock and a bond.
The architecture illustrates how developers are attempting to separate an agent’s ability to reason and act from its authority to access information or complete sensitive transactions.
Whether those protections prove sufficient for widespread financial use will depend on their performance in practice, including their ability to prevent unauthorized actions, malicious instructions and accidental disclosure of sensitive information.
Deep Financial Infrastructure
One of the most consequential aspects of Muse is its integration with existing financial technology infrastructure.
On September 8, Plaid announced that its account connectivity technology would power financial features within Muse. The integration allows consumers to connect financial accounts and use information from those accounts to receive personalized guidance, establish financial goals and take action through conversations with the agent.
This is an important development because access to reliable financial data is a prerequisite for meaningful personalized financial assistance.
A general-purpose AI model can explain how to create a household budget without knowing anything about the household. An agent connected to actual bank accounts, however, can potentially identify recurring expenses, analyze cash flow, recognize changes in spending and suggest actions based on a consumer’s financial circumstances.
The Plaid integration also illustrates why financial technology infrastructure providers could benefit from the expansion of consumer AI agents. Rather than replacing every component of the existing financial system, AI developers may increasingly depend on specialized companies to provide account connectivity, payments, identity verification and other regulated financial capabilities.
Meta’s September 24 announcement of additional Muse integrations reinforces that possibility. The company identified PayPal and Shop Pay among its expanding payment options, alongside commercial partners including Walmart, Best Buy, Expedia and Instacart
The integration of payment capabilities creates opportunities for AI agents to move beyond financial recommendations and participate directly in commercial transactions.
For banks and payment companies, the resulting competitive landscape is complicated. An agent could encourage consumers to move between providers more frequently, but the same agent might also generate additional payment volume and demand for the financial infrastructure supporting its transactions.
The distinction between financial service providers and the AI platforms that connect consumers to those providers is becoming increasingly important.
What Muse means for financial services
The financial industry has spent years deploying AI to improve internal operations, detect fraud, assess credit risk, automate customer service and support investment research.
Muse introduces a different competitive consideration: AI operating on behalf of the customer rather than exclusively on behalf of the financial institution.
A bank’s proprietary AI assistant might help a customer understand the institution’s savings products. An independent personal agent could compare those products with alternatives offered by competing institutions and help the customer move money elsewhere.
That change could affect customer acquisition, pricing, retention and profitability across several financial services sectors.
For banks, AI agents could make depositors more responsive to differences in interest rates, account fees and service quality. Consumers who previously tolerated unfavorable terms because switching banks was inconvenient might become more willing to move accounts if an agent could handle much of the administrative work.
Lenders could face similar pressures. An agent capable of comparing loan offers, reviewing disclosures and assisting with applications could make it easier for borrowers to evaluate competing mortgages, personal loans and credit products.
Insurers may face particularly significant changes in distribution. Personal agents could compare policies, identify overlapping coverage, evaluate premiums and help customers negotiate renewals. Such capabilities could improve consumer access to information while potentially reducing the value of distribution arrangements that depend on customer inertia.
These are plausible competitive developments, not established outcomes. Financial products are not interchangeable commodities, and the lowest advertised price is not necessarily the most appropriate choice for a particular consumer.
Insurance decisions involve exclusions, deductibles, claims handling and coverage adequacy. Lending decisions involve creditworthiness, contractual obligations and long-term affordability. An AI agent that optimizes a single variable could recommend a financially unfavorable transaction.
Consequently, the ability to compare products is only one component of delivering sound financial guidance.
Wealth management faces a more fundamental question
For wealth management firms, Muse raises questions about the distinction between financial information, financial planning and professional financial advice.
Many services historically delivered by financial advisors involve collecting information, organizing financial records, calculating projections, comparing alternatives and communicating recommendations.
These activities are increasingly susceptible to AI-assisted automation.
An agent with access to a household’s financial accounts could assemble a consolidated financial picture, identify excessive investment expenses, evaluate retirement savings progress and alert users to changes in their financial circumstances.
More sophisticated systems could potentially analyze asset allocation, model retirement income scenarios, identify tax-planning opportunities and coordinate routine financial administration.
Such capabilities could reduce the amount of time required to deliver basic financial planning services and place pressure on advisory fees, particularly where consumers perceive little differentiation between competing providers.
InvestmentNews reported that LPL Financial and Charles Schwab each declined more than 6% on September 22, while the S&P 500 financial sector fell approximately 2%. The publication connected the selling pressure to concerns about AI’s potential to reduce advisory revenue and disrupt brokerage economics.
The competitive implications extend beyond the possibility of replacing human advisors.
Devin Ryan, head of financial services and fintech research at Citizens, told the Wall Street Journal, as reported by InvestmentNews, that successive AI product launches have increased uncertainty about the industry’s future. Ryan also identified the potential for agents to move client assets more efficiently, reducing cash balances that brokerages can use to generate profits.
Consider an AI agent that continuously reviews a household’s investment accounts, identifies available tax-loss harvesting opportunities and recommends appropriate transfers between cash and investment products.
Even if the agent never provides comprehensive financial advice, it could influence the profitability of the institutions holding those assets.
This distinction is critical. A technology does not need to replace an entire financial services business to disrupt its economics. Automating a relatively narrow activity can materially affect profitability if that activity supports an important revenue stream.
Nevertheless, wealth management involves responsibilities that extend beyond portfolio construction and routine financial calculations.
Human advisors help clients navigate inheritance, divorce, business succession, retirement transitions, estate planning and emotionally difficult financial decisions. They also provide behavioral guidance during market volatility, coordinate with attorneys and accountants, and assume professional responsibilities that general-purpose consumer AI agents do not automatically undertake.
For complex financial situations, the value of advice may depend as much on judgment, accountability and the management of competing objectives as on the underlying calculations.
Muse therefore presents wealth management firms with both a competitive challenge and an opportunity to reconsider which services they deliver, how those services are priced and where human expertise contributes value.
Three Approaches to Financial AI
Muse’s arrival coincides with the introduction of increasingly specialized AI products for financial professionals.
On September 14, Anthropic launched Claude for Financial Advisors, a suite of connectors and workflow capabilities designed to help advisors work across the custodial, portfolio management, customer relationship management and financial planning systems they already use.
The product is intended to assist with research, client meeting preparation, documentation, follow-up activities and other administrative tasks. Anthropic emphasizes that advisors retain responsibility for reviewing and approving consequential actions.
Claude for Financial Advisors approaches the financial AI opportunity primarily from inside the advisory firm.
Rather than asking consumers to replace their existing advisor with an independent AI agent, it aims to make the advisor more productive by connecting information and automating parts of the advisory workflow.
An advisor preparing for a client review might ordinarily retrieve information from several applications, examine portfolio performance, review the client’s financial plan, consult previous meeting notes and prepare a discussion agenda.
Claude can help consolidate those activities into a coordinated workflow, potentially reducing preparation time and allowing advisors to devote more attention to clients.
Anthropic’s broader financial services strategy also includes AI agents for investment research, financial modeling, know-your-customer reviews and accounting processes.
OpenAI is pursuing a related but distinct opportunity.
On September 10, the company introduced ChatGPT for Financial Services, a specialized ChatGPT Work offering combining financial data, AI reasoning and tools for research, financial modeling and the preparation of professional materials.
Developed with input from Morgan Stanley and Evercore, the offering incorporates financial information from providers including Daloopa, PitchBook and LSEG News. It supports activities such as company research, earnings analysis, valuation modeling, acquisition screening and pitchbook preparation.
The product also incorporates enterprise security, administrative controls and mechanisms for tracing financial information to its underlying sources.
These features reflect the requirements of institutional financial work, where an apparently plausible AI-generated answer is insufficient unless analysts can verify the underlying information and explain the assumptions supporting their conclusions.
OpenAI has also introduced a separate consumer personal finance experience within ChatGPT. That offering allows eligible users to connect financial accounts, examine their financial activity and ask questions grounded in their personal financial information.
Taken together, these developments illustrate two increasingly interconnected markets for financial AI.
One market serves financial institutions and professionals, helping them improve research, operations, analysis and client service.
The other serves consumers, helping them understand their financial circumstances, evaluate alternatives and potentially undertake financial activities with less direct assistance from traditional providers.
Muse is particularly significant because it brings consumer-oriented agentic capabilities into this second market while relying on financial technology integrations to support personalized financial experiences.
Claude for Financial Advisors and ChatGPT for Financial Services, meanwhile, illustrate how AI developers are seeking to become essential components of the technology infrastructure used by financial professionals.
The competition is therefore not limited to which company develops the most capable AI model. It also concerns which companies control access to financial information, connect the relevant applications and become the preferred interface for completing financial tasks.
Financial AI’s Next Challenge
The expansion of agentic financial AI creates risks that are more consequential than those associated with ordinary conversational assistants.
A chatbot that produces an incorrect explanation of a financial product may misinform a user. An agent that acts on an incorrect explanation could initiate an unsuitable transaction, disclose confidential information or create an unintended financial obligation.
Financial institutions consequently need to distinguish between allowing AI to retrieve information, generate recommendations and execute transactions.
Each activity introduces a different level of operational and regulatory risk.
An AI agent may be capable of recommending a securities transaction, but that technical capability does not automatically establish that the provider is authorized to deliver regulated investment advice or execute trades.
Similarly, a consumer’s authorization to connect a bank account does not necessarily establish permission to initiate every transaction that an agent might consider beneficial.
Financial firms integrating with consumer agents will need to address consent, authentication, transaction limits, recordkeeping, cybersecurity and the allocation of responsibility when something goes wrong.
The risks are not hypothetical in the broader context of AI development.
Reuters reported on September 22 that Meta was testing a human concierge feature in which contractors handled some telephone calls initiated through Muse. The experiment raised internal privacy concerns about the possibility that sensitive user information could be exposed to human operators. Meta said the testing was intended to improve the feature’s safety and privacy before a wider release.
The episode illustrates a practical difficulty with agentic AI: A system presented to consumers as an automated assistant may depend on a combination of AI models, external services, human assistance and third-party infrastructure.
Financial institutions need to understand those dependencies before allowing an external agent to interact with sensitive customer information or regulated workflows.
The financial consequences of AI also extend beyond individual transactions.
The Bank for International Settlements has warned that rapid AI investment introduces new financial stability considerations, including the scale and financing of infrastructure spending. These concerns are separate from the immediate competitive pressures created by Muse, but they demonstrate that financial institutions must evaluate AI both as a technology affecting their businesses and as an investment theme affecting the broader financial system.
The future of financial services may depend on who controls the customer relationship
The most consequential question raised by Muse is not whether consumers will eventually use AI to manage their finances. That process is already underway.
The question is how financial institutions will participate when AI becomes an increasingly important intermediary between consumers and financial products.
Banks, insurers, brokerages and wealth management firms could respond by developing proprietary agents, integrating with consumer AI platforms or making their existing products and services more accessible to external AI systems.
Each approach involves trade-offs.
Proprietary agents may help institutions retain control over customer relationships, but they must compete for consumer attention against general-purpose assistants capable of coordinating activities across numerous providers.
Integrating with external agents may create new distribution opportunities, but it could also weaken direct customer relationships and increase dependence on technology companies that control the consumer interface.
Financial institutions may ultimately find themselves competing not only for individual customers but also for the recommendations and transactions generated by those customers’ AI agents.
That prospect creates opportunities for financial technology providers capable of supplying reliable data, secure connectivity, identity verification, payments and other services that agents require.
It also creates opportunities for financial advisors who use AI to improve productivity while differentiating their services through professional judgment, personalized relationships and accountability.
Muse, Claude for Financial Advisors and ChatGPT for Financial Services represent different approaches to the same underlying transformation: Financial AI is moving from generating information toward coordinating and completing work.
For consumers, that transition could reduce administrative burdens, improve access to financial information and make it easier to compare competing products and services.
For financial institutions, it could improve operating efficiency while simultaneously increasing competition, reducing switching costs and challenging established sources of revenue.
For wealth managers, it raises questions about the appropriate division of responsibilities between automated financial analysis and professional advice.
The financial stock volatility surrounding Muse reflects the uncertainty associated with that transition. Investors are attempting to evaluate the future economics of businesses whose competitive advantages were established before consumers could delegate complex activities to AI agents.
The outcome remains uncertain. Consumer adoption, regulatory requirements, technological reliability and institutional responses will determine how much of the financial services value chain ultimately changes.
What Muse has demonstrated is that agentic AI can influence the market’s expectations for financial institutions even before its long-term commercial impact becomes clear.
The next phase of financial AI will be shaped not simply by the ability of machines to analyze financial information, but by the extent to which consumers and institutions are willing to authorize those machines to act on it.
Researched by DWN Staff
