The US$1M–10M client has an advisor, pays for advice, and is not receiving it. That gap is the opportunity.
The individual is on their own: whatever the branch offers, the algorithm surfaces, or they learn themselves.
Has an advisor. Pays for advice. Does not receive it. A book of ~250 clients, under an hour each per month.
Works, but expensive, complex, and demands dedicated staff most would rather not own.
A client with 90% in CDI feels safe. In reality it is a concentrated emerging-market credit bet (one sovereign, one currency, roughly 1% of the global investable universe), indistinguishable from an EM sovereign bond fund.
The client must feel heard before they are advised. Every design decision is filtered through that one principle.
Not a questionnaire. The advisor leaves with nine inputs, captured through natural conversation.
Where it came from, where it is going
When the money must start working
How predictable the cashflow is
Any single exposure that dominates the book
The number at which they would act
Exit, inheritance, purchase ahead
Cash the portfolio must supply in 6–24 months
Regimes, structures, offshore in place
What "doing well" is measured against
Risk, liquidity, concentration, and cost & tax, run together before any recommendation. The logic behind each: §03.
Position by position, each with a thesis: its role, why this instrument, the transition path. Anything outside it escalates to the investment team.
Accept, modify, or reject. Every override is logged with the reasoning; the AI's original is kept for comparison.
Current → recommended is staged, never a jump: around illiquidity and psychological barriers, paired with the Behavioral Covenant.
Each background hides risks clients don't volunteer. The lens tells the advisor where to look, then recedes once the real numbers are in.
The longer the horizon, the more of the book starts in equity, the growth layer. A binding constraint (concentration, behavior, capacity) then pulls it back from this starting point.
The logic the recommendation is built on: how the tool reads risk, sizes the portfolio in layers, and weighs concentration. The reasoning that makes every position defensible.
What they did in 2008, March 2020, 2022
The drawdown the plan can absorb
What "well" is measured against, often CDI
What they say about risk
6–12 months of obligations, same-day liquidity.
Extends protection from months to years: NTN-B, high-grade credit.
The residual, globally diversified by default.
Bounded, only on an active investment-team view.
Documented and raised at the next touchpoint. Baseline unchanged.
Allocation modified to reflect the exposure. Advisor action logged.
The portfolio is built around the position, not despite it.
Every position is judged net of fees and tax. A lower headline yield with a cleaner structure can keep more than a higher one that leaks.
AI does what scales. The human does what matters. The split is deliberate.
150+ clients, each actually advised, not fewer. Protecting quality caps today's advisor near 150; the AI removes the manual work behind that ceiling, so 150 × 4 conversations at ~1.5h each ≈ 900 of an advisor's ~1,700 yearly hours.
A byproduct, not a module. Every thesis, override, and version is already documented by the workflow.
Improves with use. More clients → better-trained AI → better outcomes. A moat that compounds.
Succession, alternatives, life-encompassing advisory. Same tool, more service density.
Where the pain is sharpest. AUM-based, advisor-led, and where the model is proven.
AI-first, subscription / SaaS, a personal financial operating system.