Imagine finishing a client meeting and finding a summary ready before you have closed your laptop. That is useful. You still need to check what it says, confirm which decisions the client actually made, assign the follow-up and make sure the next conversation starts with the right information.
The summary took less time. How much of the job is finished?
Faster tasks and a full calendar can coexist
In a September 2026 release, Vanguard reported that 72% of surveyed advisors wanted more time for prospecting and deepening existing client relationships. The research covered 549 U.S. advisors surveyed in July.1
AssetMark published a different finding that same week. Among AI adopters in its survey, more than half reported saving at least four hours a week. Its broader sample included 400 U.S. advisors, surveyed between June 24 and July 7, and excluded existing AssetMark clients.2
These are different groups answering different questions. They do not prove that AI savings are disappearing or that technology has failed.
An advisor can save time on recurring tasks and still want more time with clients. Perhaps the practice has grown. Perhaps the advisor now prepares more thoroughly. Perhaps the saved minutes arrive in small gaps that never become a usable appointment. Those are possible explanations a firm can examine, not conclusions established by these surveys.
There is also a measurement problem. Self-reported hours saved describe the respondent's experience. They do not tell us how much review remained, whether someone else absorbed the work or what happened to the recovered time.
Count the work around the output
Take the meeting summary in our opening example. Suppose it captures the discussion well and produces a sensible list of next steps. The advisor may still have to check a figure against the source document, distinguish a suggestion from an agreed action and decide who should contact the client.
Then someone must put those actions where the team will see them. A task with no owner may sit untouched. An approved follow-up that never reaches the client has not completed the promise made in the meeting.
Follow one review meeting through preparation, the conversation and the work afterward. Notice each point where someone searches for information, copies it into another system, checks it or asks a colleague what happened. Include the time spent correcting mistakes.
Be careful about treating every check as waste. An advisor reviewing a recommendation or correcting an inaccurate summary is doing necessary work. Removing that check to make the time-savings number larger would be a poor trade.
The useful distinction is between professional review and avoidable reconstruction. Reading a draft to confirm that it reflects the meeting has a purpose. Reopening several records because the draft gives no indication of where a fact came from is a different kind of work.
Decide what the time is for
A practice needs a view on where recovered time should go. Otherwise, the next urgent request can become the default answer.
Growth is one possible goal. A firm might want room for more introductory meetings without weakening service for existing clients. Another practice may want to clear overdue follow-up or prepare more carefully for complicated family conversations. An advisor with a crowded calendar may simply want space for the unexpected.
Spare time in a service business can have a purpose. A client calls with a problem that will not wait until the next scheduled review. An appointment takes longer because something important has changed.
Choose an observable outcome before judging the tool. If the goal is better follow-through, look at whether agreed actions reach completion. If the goal is more client contact, look at whether those conversations actually happen. A busier calendar, on its own, says little about either result.
Quality also belongs in the assessment. A practice may reasonably keep a tool because it produces a more complete first draft, even if the review process takes about the same time. It should describe that as a quality improvement and evaluate it on that basis.
Keep client context connected
For a review meeting, relevant context might include the client's current goals, an unanswered question from the last conversation and the document behind a figure. If those details disappear between preparation and follow-up, someone has to find them again.
A useful connection between systems should help the next person understand what matters and what needs to happen without rebuilding the client story.
At Adrima, we're building around this continuity: bringing client goals, recent activity, open questions and supporting information into preparation briefs, and connecting transcript review, notes, tasks and draft follow-up. These capabilities are in development, with advisor review built into the intended workflow. Our intended approach works alongside existing CRM, portfolio, planning and document systems, with connections dependent on the systems available.3
For any product taking this approach, the evaluation should follow an actual client task. Can the advisor check the underlying information? Does the next action reach the person responsible? What still needs to be entered again?
Follow one workflow all the way through
Someone needs responsibility for that evaluation. Otherwise, each team member can improve an individual task while nobody checks whether the overall process became easier.
Cerulli's September research with Vista Equity Partners highlighted ownership, governance, training and measurement in its assessment of firms' AI maturity. The study covered 68 RIA firms and more than 20 interviews. Its average participating firm had $18 billion in assets, so a small practice should be cautious about treating it as a description of its own circumstances. The findings also do not establish that a particular operating practice caused better results.4
A firm can still start with a modest exercise. Choose one repeated process, such as preparing for and completing a client review. Write down its starting point and what counts as finished. Then observe several reasonably comparable cycles, including an awkward one that needs corrections or extra information.
Keep four questions in view:
- How much staff effort did the complete process take, including preparation, review and corrections?
- How long did the client wait between the initial conversation and the completed follow-up?
- Where did information need to be found again or entered twice?
- What did the practice actually do with any time it recovered?
These observations will not produce a universal verdict on AI. They can show whether a particular change helps this team with this job. Keep the exceptions visible. A polished routine case can hide the work required when a document is missing or a client changes direction.
Then make a specific decision. Adjust the handoff, clarify the owner, improve the information available for review or reconsider the tool. Judge the result against the purpose you chose.
For the next client review, follow the work until the promised action is complete. That is where the practice can see whether a faster task has made room for better service.
Footnotes
-
Vanguard, survey release on advisor technology and client time, September 15, 2026. U.S. survey, July 2026. ↩
-
AssetMark, Advisor Insights Report on AI, September 15, 2026. Savings are self-reported by AI adopters, a subset of the 400-advisor U.S. sample. ↩
-
Adrima Editorial, Our first Adrima update: what we're building, September 12, 2026. Describes work in progress; accessed September 19, 2026. ↩
-
Cerulli Associates, Advisor headcount set to grow as AI expands capacity, September 9, 2026. Research with Vista Equity Partners, May through July 2026. ↩