Who Fills the Seat
Updated: 2 days ago

A bank now lets AI act on your account, at your own risk. What sits beside you should be formed for you and owned by you, not simply available by subscription.
On 25 August 2026, Scalable Capital, a Munich bank and broker with more than a million clients and over €60 billion on its platform, launched what it calls Agentic Investing. Clients can now connect general-purpose AI assistants, ChatGPT, Claude and Grok, to their accounts. Through them they can trade stocks, ETFs and derivatives, set up savings plans, adjust limit orders, run portfolio analyses and risk assessments, compare their holdings with model portfolios and receive a daily newsletter built from their own positions. Every trade must be explicitly confirmed, with its cost information and key information document. Logins need two-factor authentication, money cannot be deposited or withdrawn through the assistant, and the feature can be switched off at any time. The bank’s co-founder, Erik Podzuweit, called it “the greatest technological shift in financial technology since internet banking.”
The same release contains a sentence that matters more. The AI’s outputs, recommendations and transactions “do not originate from Scalable Capital, are not investment advice, and are executed at the investor’s own risk,” and the assistants themselves “operate independently and outside Scalable’s control.”
As such, it is expected. Millions already use these assistants every day, and connecting them to the place where money sits was the obvious next step. The design is careful. Scalable is a retail broker, not a private bank, but retail is often where a habit forms first. The principal who already asks an assistant about a contract or a journey will soon be offered the same connection to an account, if not by this bank then by the next.
Now take it from the chair of the principal — your chair. Through this series the seat on your side of the table has stood empty: the firm kept the minutes, held the map, set the dial and charted the course. Here, for the first time, something sits in that seat. But look at what it is. It is not yours. It is the assistant available to anyone with a subscription, the same one your neighbour uses and your adviser’s firm may use, formed by no one for you in particular. And the bank has drawn its line around it: not from us, not advice, at your own risk, outside our control. The seat is filled, and the whole responsibility for what fills it has been placed on you.
Then ask what it knows. It sees one account. It does not see your other holdings, the trust, the property, the company shares, the obligations to family, or the purpose the money is meant to serve. The bank’s chief product officer, Alexander Seipp, said the feature “certainly creates a level-playing field” and that it takes “just a few prompts.” Both are true of access. A study by Elm Wealth, cited when the launch was reported, shows what a few prompts leave out. Tested on historical news, Claude chose better than human players in 76 per cent of sessions, yet the models were poor at deciding how much to stake, taking “too much risk of a catastrophic loss of capital.” What to buy can be read from the market. How much depends on everything the assistant cannot see.
The confirmation step does not close that gap. It asks you to approve the trade, with its costs and documents, and it protects the transaction. It does not ask whether the reasoning behind the trade knew enough about you to be right. Approving a suggestion from an assistant that sees a fraction of your affairs is not the same as making a decision. Seipp himself put it honestly: whether the AI will deliver good results “remains to be seen.”
From your chair, the principal’s, the questions are these. If you connected an assistant to your accounts tomorrow, what would it know about you beyond the balance? Who decided how much risk is right for you, you or the assistant’s defaults? When you confirm a trade, are you confirming a decision or a suggestion? And whose is the assistant in the seat beside you?
The first answer is practical. Before connecting any assistant to any account, write down what it ought to know and what it must never do: the purpose of the money, the limits you hold, the parts of your picture that sit elsewhere, the decisions that remain yours alone. Without that, it acts on a fraction of you.
The deeper answer is to fill the seat with something that is truly yours. Not the general assistant as it arrives, but the same technology formed around your affairs, your intentions and your limits, and owned by you in every respect: its instructions, its material, its record of your reasoning. Such a Bespoke AI Confidante can sit beside you when a bank opens its doors to AI, and read every suggestion against the whole of what you hold and what you mean to do with it.
That is what SMA Crown Confidential forms. Not another adviser at the table, but an intelligence layer on your side of it: a Bespoke AI Confidante running on the model of your choice and formed around your own intentions, so that your judgement stays independent of every platform that now invites an assistant to act on your accounts. The banks have opened the seat. The question is who fills it.
Seraphina Michelina Evelyna Bogomolova-Huotelin (nee von Prittwitz)
Founder & CEO of SMA Crown Confidential
Bespoke AI Intelligence for Private Decision-Makers
.jpeg)



Comments