SMA Crown Confidential Positioning Note: Tandems Hamilton
Updated: 4 days ago

Occasion: Tandems launched "Hamilton," an assistant built into its "AI-native WealthOS" and marketed as handling up to 80% of a financial advisor's daily tasks, including client reviews and proactive client communication (confirmed as launched; an interested party, with all performance claims vendor-asserted). Its stated controls are taken at face value: a skill, a model or all AI can be shut off in real time, every AI-assisted interaction is logged, and prohibited uses are blocked. The 80% figure, the reliability claims and the customer and scale figures are not used as fact.
Position occasioned: architectural origin, turned on where each party sits: the adviser sits with the client, and Hamilton sits with the adviser. Hamilton attends to one client at a time, but it knows that client through the institution's file, which is formed for what the institution needs to know. Lit from the client's world.
This week Tandems launched Hamilton, an AI assistant built into what it calls its AI-native wealth operating system, which it says can handle up to 80% of a financial advisor's daily tasks: client reviews, meeting preparation, proactive communication with clients. The release is specific about its controls. A single skill, a model or all of the AI can be switched off in real time, every AI-assisted interaction is logged, and prohibited uses are blocked. We take that at face value. It is how a serious firm would want to deploy such a tool.
It would be easy to read Hamilton as a tool for the client. It reviews the client's affairs, prepares the meeting, writes to the client between meetings. But it does not sit with the client. The adviser sits with the client, and Hamilton sits with the adviser. An adviser has many clients and meets them one at a time, so in each session Hamilton attends to one person, gathering what the institution knows about that client and putting it in front of the adviser.
What the institution knows is the question. A client's file is formed by what the institution needs in order to advise, serve and supervise: holdings, objectives as recorded, risk profile, suitability, the notes of past meetings. It is a legitimate record, built for the institution's purposes. It is not the whole of the client's life: their position in their family, what the wealth is for, what they are carrying that no form asks about. Hamilton can make the adviser faster and better prepared on everything the file contains. It cannot make the file contain what it was never formed to hold.
So the help runs in a fixed order: the institution's record first, the adviser's preparation second, and the client third, met through both. This is not a judgement on anyone's intent. The institution is a counterparty, not an opponent, and a well-prepared adviser can serve a client well. It is simply the order in which the chain is built.
A Bespoke AI Confidante sits at the other end of that chain. It sits with the client, and it is formed from the client: their character, their relationship to what they hold, the questions they keep returning to, not the file an institution keeps on them. That is why no institutional version of it can exist. The institution can build tools that know its file on you very well. It cannot build one that was formed from you.
Seraphina Michelina Evelyna Bogomolova-Huotelin (nee von Prittwitz)
Founder & CEO of SMA Crown Confidential
Bespoke AI Intelligence for Private Decision-Makers
.jpeg)



Comments