Asset Managers Urged to Use AI to Prevent Operational Errors

Asset managers could get more from artificial intelligence by using it to spot and prevent operational problems before they affect trades, valuations or fund decisions. A framework issued by the Investment Company Institute on October 29, 2026, argues that AI should help firms see activity across disconnected systems and give staff time to respond while choices remain open.

The proposal, published by ICI’s Financial Innovation Advisory Committee (FIAC), is called AI and the Future of Asset Management Operations: Considerations for an Intelligent Operating Model. It sets out three aims: prevent recurring errors, provide a broader real-time view of business activity, and identify emerging risks early enough for employees to act.

Connecting information across operations

The case for the model starts with a mismatch: asset management has become more complex, while operational teams still spend substantial time assembling information held in separate systems. That work grows as firms add products and distribution channels. Tighter settlement windows under T+1 rules add further pressure to identify missing or inconsistent details before they cause delays.

ICI’s Andrew Kayiira, Director, Technology & Financial Innovation, and R.J. Rondini, Senior Director, Securities Operations, described the potential in a viewpoint accompanying the paper: “Imagine identifying and correcting a settlement problem before sending the trade details out the door, or giving fund professionals a clearer picture of emerging risks while there's still time to act.” The value would come from bringing relevant information together sooner, not simply processing the same investigation faster after a problem surfaces.

A large shareholder redemption illustrates why that distinction matters. Decisions about liquidity can depend on information held by trading and shareholder-servicing teams. In a manual process, staff may have to gather that information as events unfold. An AI system could assemble the available picture automatically, giving decision-makers more context when they need to weigh their options.

Valuation offers another example. If a security’s price appears questionable before a fund’s net asset value calculation, a system could collect relevant information before the calculation runs. Specialists would then have more time to assess the price, rather than first confronting the issue after it has affected the fund’s NAV. In both cases, the proposed role for AI is to bring forward the point at which people can apply judgment.

Start with a recurring problem

ICI does not present the model as a call for firms to rebuild their operations all at once. It recommends choosing a recurring problem and testing whether AI improves the current process. Measures of progress might include fewer settlement exceptions, less investigative time for each incident, or the ability to handle more work without a proportional increase in staffing.

This kind of bounded trial can also show where data connections break down and where employees need to review a system’s work. Those lessons matter before firms attempt more ambitious applications, including agentic AI systems that can carry out multi-step tasks with limited human intervention. The paper presents incremental results as a foundation for that next stage, rather than treating greater autonomy as the starting point.

The approach tracks a wider change in how wealth and asset managers view AI. InvestmentNews has reported that firms are increasingly considering AI as part of core operating infrastructure, with potential uses extending beyond cost reduction. ICI’s framework directs attention toward the middle and back office, where fragmented information and slow escalation can shape the experience of investors and the workload of employees.

Adoption remains uneven

Evidence cited in the paper points to a gap between AI use in investment work and use in operations. A Mercer survey of asset managers in 2026 found that 55 percent had integrated AI into at least one investment process. Adoption was uneven and concentrated largely in front-office activities such as research and idea generation. The framework’s focus on operational work would extend the discussion to functions where unresolved errors can affect investors directly.

The paper also places its argument alongside a challenge from the SEC’s Division of Investment Management. Director Brian Daly, speaking at a Stanford roundtable on October 16, 2026, urged investment advisors and fund sponsors to “enthusiastically innovate in a responsible and thoughtful manner.” ICI casts that challenge as a reason for the industry to take the lead: “Director Daly has challenged the industry to lead,” the paper states. “We should answer that challenge with a fundamentally better way to operate.”

For advisory firms and asset managers, that means implementation choices will matter alongside the technology itself. The source describes AI use in valuation, compliance workflows and trade execution as areas under observation, and argues that firms need governance structures that make responsibility clear. Connected information may help surface a problem, but employees still need to understand how a system reached its conclusion and who is accountable for acting on it.

Human judgment remains central

ICI frames AI as a support layer for experienced professionals, not a substitute for their expertise. The intended benefit is more time and better information for decisions that remain difficult or context-dependent. That boundary is also relevant to advisors and wealth managers considering how much responsibility to assign to AI in client-facing or portfolio work.

InvestmentNews coverage has found that approximately 95 percent of financial services professionals plan to increase AI investment over the next three years, while many still draw clear lines around human accountability. ICI’s model reflects that tension: wider use depends on connecting information and identifying risks earlier, while keeping people responsible for consequential decisions. The practical test will be whether firms can show that those systems reduce operational friction and give staff a clearer, earlier basis for action.