Autonomous wealth management could reduce costs and widen access to investment services, but adoption will depend on data quality, regulatory oversight and accountability for software-driven decisions.
An investment policy statement defines a client's financial objectives, risk tolerance and investment constraints. Andrew Ang, a finance academic and former BlackRock executive, has proposed applying this framework to autonomous investment systems. In a paper co-authored with Nazym Azimbayev and Andrey Kim, Ang describes a system in which approximately 50 specialized artificial intelligence agents construct portfolios, compare strategies and evaluate decisions.
The agents operate within a defined investment mandate, shifting the human role toward setting objectives and overseeing decisions. The research proposes an architecture, not proven performance in live markets.
From automated portfolios to autonomous decisions
Wealth management automation has progressed from portfolio maintenance toward systems capable of making investment decisions. Robo-advisors initially used client questionnaires to assign model portfolios and rebalance them periodically. Statista projects U.S. robo-advisor assets at approximately $1.6 trillion in 2026. However, rule-based rebalancing differs significantly from delegating investment decisions to AI.
AI subsequently entered advisory workflows, supporting research, financial planning and portfolio analysis while human advisors retained decision-making authority. Agentic AI extends these capabilities by enabling specialized systems to generate investment assumptions, compare portfolio-construction methods and make decisions within an established mandate. Its effectiveness depends on reliable data, consistent outputs and controls over the actions agents can take.
The commercial case for autonomy
Cost reduction is a major incentive for autonomous wealth management. Human advisors assess client circumstances, investment objectives and portfolio performance, making personalized services expensive for smaller accounts. Automation could reduce manual work in rebalancing, monitoring and tax-related management, enabling firms to serve more clients without proportionately increasing staffing.
However, autonomous systems require investment in data infrastructure, model validation, cybersecurity, monitoring and compliance. These expenses could offset savings, particularly for firms managing complex platforms.
The existing robo-advisory market does not establish that investors will accept greater autonomy. Firms must demonstrate that expanded authority delivers sufficient value to justify its risks.
Three models of AI-led wealth management
Delegated authority determines the operating requirements and accountability of AI-based investment services.
Advisor-assisted management keeps investment decisions with human advisors. AI prepares research and recommendations for review and approval. Firms must maintain reliable data and records of recommendations and human decisions.
Hybrid management allows AI to execute routine actions within predefined thresholds, requiring human approval for decisions outside those limits. This model needs clear permissions, client consent, escalation procedures and alerts when portfolios approach established boundaries.
Fully autonomous management authorizes systems to select and execute trades within a client's investment mandate. It requires continuous monitoring, auditable records, informed consent and procedures for intervention when systems fail or breach their limits.
Data quality and investor trust remain barriers
Institutional readiness constrains adoption. Many asset managers operate legacy portfolio, custody and client-management systems not designed for autonomous agents. Fragmented data can undermine model outputs, while shortages of professionals combining investment and machine-learning expertise complicate implementation.
FE fundinfo's June 2026 survey of 200 senior asset managers across the United Kingdom, Luxembourg and Switzerland found that 64% believed AI would deliver meaningful value only after firms improved underlying data quality and structure. Just 36% described their AI use as broad or fully integrated. The findings suggest that scaling AI requires operational improvements alongside new technology.
Investor confidence presents another obstacle. Janus Henderson's 2026 survey of 1,000 affluent and high-net-worth U.S. investors, conducted with 8 Acre Perspective, found that 72% cited insufficient trust in AI-driven recommendations as a barrier. Additionally, 85% said their advisor was ultimately responsible for AI-generated advice or materials, while 79% would be upset if their advisor used AI without disclosure.
Although the survey examined financial guidance rather than fully autonomous mandates, it highlights the importance of transparency and accountability. Investors may accept AI as an advisory tool without accepting complete delegation of investment decisions.
Regulation and systemic risk
Wider adoption could create risks beyond individual portfolios. The Bank of England's Financial Policy Committee has warned that AI could contribute to herding if firms rely on similar data, models or investment signals. Ang's architecture incorporates competing portfolio-construction methods, but using multiple agents does not guarantee independent decisions or prevent correlated trading. Its effectiveness remains unproven in live markets.
Regulatory expectations are evolving. In the United States, fiduciary obligations remain relevant when financial professionals use AI to provide investment advice. In Europe, requirements for wealth managers depend on the systems and activities involved, including applicable implementation timelines for high-risk AI obligations.
Singapore has adopted a more explicit approach to financial-sector AI governance. On Oct. 7, 2026, the Monetary Authority of Singapore issued AI risk-management guidelines for financial institutions, with implementation beginning Oct. 7, 2027. The guidelines set risk-proportionate expectations for oversight, risk management and controls across the AI lifecycle.
Accountability remains with the deploying firm
Autonomous wealth management could expand access to investment services and reduce routine portfolio-management costs. However, delegating decisions to software does not remove institutional responsibility. Firms must define delegated authority, monitor decisions, maintain auditable records and intervene when systems exceed their mandates.
The central question is whether institutions can demonstrate that autonomous decisions remain consistent with client objectives, risk limits and regulatory obligations. Firms that establish effective controls and disclose AI's role in investment decisions will be better positioned to expand adoption. Without these safeguards, autonomy could increase investment and operational risks without delivering sustainable benefits to clients.
