Essay & podcast · Agentic Wealth
Agentic wealth: unlocking the agency of advisors
A client pays an advisory fee, and the advisor keeps a shrinking share of it — the rest goes to layers. Agentic AI doesn't replace the advisor; it replaces the layers between the advisor and their clients, and hands back the time, the economics, and the agency.
Start with where the money goes. A client pays an advisory fee, and the advisor — the person doing the actual work of judgment and relationship — keeps a shrinking share of it. The rest is claimed by layers: model marketplaces charging ongoing basis points for allocation that is, for evidence-based practices, largely settled; an underlying fund expense layer, charged whether or not it's earning its keep; a dozen software subscriptions that don't know the advisor's clients or voice; compliance support sold by the document; platforms taking margin for scale that mostly serves the platform. The industry calls this infrastructure. Some of it is. Much of it is the same problem I've written about elsewhere wearing different clothes — accreted complexity, paid for in the one currency an advisor cannot print more of: time.
My working view is that agentic AI changes the economics of those layers, and changes them in the advisor's favor. Not by replacing the advisor — that is the industry's current fantasy, and I think it has the direction backwards — but by replacing the layers between the advisor and their clients. No rail replicates taste, judgment under uncertainty, or the read on a client that comes from years of sitting across the table from them — that stays exactly where it belongs.
The line between what stays human and what becomes an agent is easier to draw than the debate suggests. Work that is testable, repeatable, and settled belongs on rails: Marketing-Rule screens run at draft time, allocation mechanics executed against a written methodology, meeting preparation assembled from records the practice already holds, the planning opportunities a complex household's own paperwork already contains but a busy practice might miss, the twentieth quarterly letter that says what the nineteenth said with new numbers. Work that requires knowing a family, weighing a trade-off, or sitting with uncertainty belongs to the advisor — and always will, because that is the part clients are actually paying for.
A practice that draws the line there gets both halves right. The rails grow more reliable with use. The advisor's hours concentrate where they were always most valuable. And the economics begin to move: when drafting, screening, and preparation run on software the practice controls, the layers that used to bill for those functions have to justify themselves against a marginal cost near zero. Most cannot.
We build for one methodology deliberately. Evidence-based investing — broad diversification, cost discipline, tax awareness, no market timing, no certainty language — is where the research record is strongest, and where advisors already share enough philosophy to share rails. That is an intentional narrowing of who we serve, not a judgment that other approaches cannot be practiced well. A large population of advisors already practices this way, on rails designed decades ago. They deserve better ones.
What we do not claim: to know how fast this happens, which firms adapt, or where the models plateau. I am trained to respect what I don't know and the uncertainty the future holds. The thesis is about direction, not timing — and the way to test it is to build in the open and let advisors judge the tools by their work. That is what we are doing at williamfrank.ai: free tools first — an agent that drafts in your voice and remains your IP, a compliance screen that returns specific edits rather than a black-box no, a model critique that shows its assumptions. Behind them, for advisors who want to go all the way, WealthFactor — where a small group of evidence-based advisors practices on these rails while keeping their own voice, book, and clients.
Your voice. Your judgment. Your clients. The rails should work for the advisor — not the other way around.