What if the next order on your investment account wasn’t typed by you, but by an AI agent acting on your behalf? Scalable Capital may soon find answers to that question with its new agentic trading solution.
German neobroker Scalable Capital announced on August 25, 2026, that its customers can now connect AI assistants such as ChatGPT and Claude directly to their brokerage accounts. Once linked, these AI agents can analyse markets, monitor a customer’s holdings, and place trades with the customer’s approval. Scalable Capital, which serves millions of retail investors across Europe, says this makes it the first bank on the continent to open its platform this way. The company said account holders could now use major AI platforms like ChatGPT and Claude to place trades and analyse their portfolios, making an early step toward broader AI adoption among everyday investors.
What “AI agent trading” is
For readers unfamiliar with the jargon: an “AI agent” is a chatbot-style assistant, like ChatGPT (made by OpenAI) or Claude (made by Anthropic), that can do more than just chat. It can connect to outside services and take real actions, such as checking a bank balance or, in this case, sending a stock order. This is part of a wider trend called “agentic AI,” where assistants move from giving advice to actually carrying out tasks. Under the new setup, called Agentic Investing, a customer could simply type a request like “check my portfolio’s diversification” or “buy €500 of a specific ETF,” and the AI would carry out the analysis or route the trade request to Scalable Capital, pending the customer’s final sign-off.
Trading with AI may be convenient, but who’s accountable?
Despite the convenience, AI agent trading is the new grey zone of responsibility. Scalable Capital is careful to note that these AI tools operate independently of the company, and that anything an AI agent says or does isn’t investment advice from Scalable Capital itself. That raises a string of unresolved questions for everyday investors: Who is responsible if the AI misreads a portfolio and suggests a bad trade? How does a broker know a trade request genuinely reflects what the customer wanted, rather than a chatbot’s own interpretation? And could someone trick or “prompt” an AI agent into making an unwanted trade?
There is similar debate now regarding agentic payments and AI bot actions in general. Earlier this month, Nvidia, Cisco, CrowdStrike, Cloudflare, Microsoft, IBM, Hugging Face, Databricks, Salesforce, SpaceXAI, and hundred more tech firms united under the Open Secure AI Alliance (OSAA) SAFE Framework umbrellf, aiming to standardize reporting for cases when AI agent or an automated system makes unauthorized, unintended or otherwise unsafe actions.
Scalable Capital says trades still require the account holder’s explicit approval and multiple security checks, but the underlying accountability questions remain largely untested by regulators. Therefore, Scalable’s co-CEO Erik Podzuweit described the launch as a “first step” ahead of wider AI adoption, acknowledging that many customers may still be uneasy about letting a chatbot near their money.
