87% of firms use or pilot AI; 6% run agentic tools inside workflows

The work AI has reliably taken over inside a wealth management firm is real, and it is narrow. Meetings get prepped faster, client emails go out cleaner, prospect research takes less time; WealthManagement.com assembled exactly that list in a column asking why the technology isn't doing the real work. Nothing on it reaches the operational and administrative load that fills an advisor's day, which is why the outlet's answer to its own headline turns out to be about budgets rather than models.

Firms have spent recent years buying AI tools, training staff on them, and telling employees and clients alike that they are now an "AI-forward organization." As those budgets grew, so did the scrutiny from CEOs, CFOs and private equity investors who want to know what the spending returns. For most firms, the outlet writes, the honest answer is less than expected.

Three figures size the gap. The 2026 WealthStack Study found 87% of firms use or pilot AI, and Deloitte found 6% use agentic tools to complete tasks inside workflows rather than assist with them. F2 Strategy's Q2 2026 Trend Report, a survey of 40 leading RIAs, wealth management firms and broker-dealers representing $8.6 trillion in assets, found a very weak correlation between what firms spend on AI technology and meaningful, measurable value to the business. Set side by side, the first two describe an industry that has bought in almost completely and a small fraction that has gotten past the front door.

The time data puts a cost on the shortfall. Kitces Research, cited in the piece, found that top-performing advisors spend roughly 30% to 35% of their time meeting directly with clients, and that most advisors fall well short of even that. The reason the outlet gives is straightforward: operational and administrative work claims most of the day, and AI investment has yet to reclaim that time in any meaningful way.

The outlet's diagnosis is a purchasing problem rather than a model problem, phrased as firms having bought AI that sits alongside the work instead of inside it. Its image is a six-lane highway capable of moving enormous amounts of work, with most firms stuck on the on-ramp behind a jam of data they cannot reach. What put them there is the legacy stack: disconnected layers of systems that the article compares to a Rube Goldberg machine, an overly complicated apparatus assembled to perform a simple task in indirect ways.

Follow the handoffs and the budget question changes shape. In a stack of disconnected layers, any workflow that crosses a system boundary — account opening that starts in the CRM and finishes in reporting, a review that draws on the planning tool and lands in a meeting note — needs a person to move the data across. Those handoffs are where the promised hours were supposed to come from. That reading is mine rather than the article's; what the article establishes is where the failure sits, in data that stays unreachable and work that stays outside the tool.

For an advisor weighing a next purchase, the audit that follows is cheap and unglamorous. List the recurring workflows that cross a system boundary, count the human handoffs inside each one, and price the tool against the handoff rather than against the demo. A connector that moves client data among planning, reporting and CRM without anyone retyping a field addresses the constraint the F2 Strategy data describes. Another assistant that drafts a client email does not, which is one plausible account of how 87% of firms ended up with faster emails and the same crowded afternoon.

The adoption question is settled and the integration question is not. If the binding constraint is data that does not travel between layers, then the return on AI spending is set by how much of a firm's stack it has wired together rather than by which product it licenses. The number worth watching next is not the adoption rate, sitting at 87%, but how many of those firms can name a workflow where nobody touches the data by hand.