The AI race moves to what happens after the meeting

Salesforce has planted its flag on a particular question about what artificial intelligence is for in a wealth practice: not the notes, not the meeting itself, but the follow-through that used to die in an inbox, and its new Agentic Advisor suite is built to turn a meeting's output into a prioritized set of actions — the tasks, deadlines, and client commitments that a sharp associate would catch and a busy one might not. The product is the news. The argument underneath it is larger, and it is the one that matters to anyone trying to grow a book, because the AI gap in wealth management has moved out of the back office and into the part of the practice that touches clients, where the referral and the consolidation dollar get decided.

For three years the advisor-tech race has been run on the back half of the meeting — transcription, summaries, a drafted follow-up note — useful work and largely revenue-neutral, saving an hour and rarely moving a dollar, while Agentic Advisor aims at the half that does. Any substantive client meeting, with a prospect or a longtime client, throws off a dozen loose ends: a statement to pull, a beneficiary form to update, a call promised to the client's accountant, a question about the concentrated position that got deferred past the hour. The practice that closes those loops in the fortnight after the meeting is the practice that gets the next introduction, and the one that doesn't is the one whose client cannot quite say why a second account got opened somewhere else.

The referral economics are worth stating plainly, because they are what turn this into a growth play rather than a productivity upgrade: referrals and asset consolidation are the two cheapest dollars in an advisory practice, with no acquisition cost, no marketing spend, and no campaign that has to clear compliance review. Both arrive from clients who have noticed that their advisor does what was said in the room, roughly when it was said. The assets that consolidate are the ones sitting at an old custodian because a previous conversation produced a to-do that never got completed, and the introduction comes from a client for whom referring a friend felt safe because the last three requests landed on time. Follow-through, read this way, is not a service nicety bolted onto a good relationship but the mechanism that turns an existing client into a larger one, and it runs on capacity most practices do not have.

That capacity gap is why 73% of the RIA firms in PWD's tracking say they plan to hire junior advisors, with 15% planning to add compliance staff, and read as a bet the hiring pattern is revealing. Firms are staffing the human layer that AI is supposed to amplify, adding the people who will run the follow-through and, in a smaller share of cases, the people who will review it, which makes a junior advisor the human version of what Agentic Advisor is selling and every hire a wager that the volume of client commitments is climbing faster than the software can absorb it. The same bet shows up in the spending, where Mariner is putting $175 million into bots at $250,000 per bot over five years, less a software line item than a headcount purchase — capacity bought in the shape of a machine rather than a person. Set the two figures side by side and the 73% stops reading as an AI adoption story and starts reading as an AI conversion problem: the industry is buying follow-through capacity, human and machine both, faster than it is turning that capacity into client outcomes, and it is buying it ahead of the workflows that would aim it at anything in particular.

The mechanics are unglamorous, and that is the point. A meeting ends; the commitments it produced get captured the same hour rather than the next morning; each gets a name and a date; and the list gets walked at the Monday stand-up the way a pipeline gets walked. The software earns its keep by removing the step where a busy advisor intended to write those commitments down and didn't — the same failure that produces the dormant account at the old custodian. What no suite can do is decide which commitments matter, or get an advisor to trust the list enough to run the week by it, a judgment that stays with the practice, which is why the junior hires and the subscription are complements. Someone still has to own the list, and someone still has to notice when an owner doesn't.

Follow-through only compounds when it can be audited, and that is the flank of this race that gets less attention than the build-out. Conquest dropped "Planning" from its positioning and bet instead on a white-label advice engine whose compliance argument is the audit trail behind SAM — the record of what the firm recommended, and when — while Hamachi went the other direction and launched a patent-pending masking layer meant to keep client data out of large language models altogether, two answers to the same question. Once a system is drafting the follow-up, the task list, and the recommendation that rides with them, the practice has to be able to show what happened and where the client's data traveled — before the exam, or before the client's attorney asks for the file, not after — and that unglamorous half of the follow-through race decides which firms get to keep the ground they gain.

Cerulli counts 35% of advisers retiring within a decade, which turns a follow-through problem into a succession problem, and that clock makes it urgent rather than merely interesting. A large share of the relationships now being handed to the next generation are held by advisors who have run them on memory and personal diligence for thirty years, and when they step back, the institutional record of what each client was promised in each meeting walks out with them unless it lives somewhere a system can act on. Agentic Advisor's premise — meeting output becomes a prioritized action list — is, under that reading, a succession instrument wearing a productivity label. It is the firm's memory of what it owes its clients, written down and routed to whoever is still at the desk.

The open question is adoption, and it is narrower than the suite's marketing implies. Salesforce can build the thing; the variable is whether a practice can wire it into how it already works, and the integration is a project, not a download — a CRM clean enough to route an action to a named owner, data-custody rules strict enough to survive an exam, junior staff trained to work a list rather than admire it. The firms that can absorb that project are the Mariner-scale ones and the ones already on their way there. Everyone else buys the note-taking tier, watches referrals hold flat, and tells itself that AI didn't deliver, a verdict that would be wrong and expensive and the one the middle of the market is most likely to reach, because the parts of follow-through that generate revenue are the parts that demand the most unglamorous groundwork first.

The AI gap is no longer a question of whether firms adopt the technology; it is a question of whether they point it at the meeting's aftermath, where the growth is, instead of the meeting's minutes. Agentic Advisor puts that choice in front of every practice running a mainstream CRM, and the firms that treat meeting output as a pipeline — every commitment an open task, every task owned, every owner measured — will compound referrals and consolidate outside assets while smaller competitors are still comparing transcription vendors, while the firms that don't will keep hiring juniors to do by hand what a rival has routed to a machine, and they will feel that cost every month it continues. More telling than how many practices pilot the suite is how many of them leave a Monday meeting with an action list that has owners and dates by the end of the week, and how many of the junior advisors they are hiring were trained to clear it.