A major German bank just let Claude and ChatGPT trade for customers. In one test, Claude beat human traders 76% of the time—but there’s a catch

Scalable Capital is launching “Agentic Investing,” opening its platform to AI assistants like ChatGPT, Claude and Grok so European investors can analyze portfolios and place trades via prompts using MCP. The company reported €60B+ in client assets and 1M+ customers, and said trades/savings plans require user approval with no AI ability to withdraw or make payments. A June Elm Wealth test suggests AI may predict market direction better than humans (Claude 76% of ~200 sessions; ChatGPT 63%) but tends to take too much risk on position sizing, making real risk management a key open question.

Analysis

This is more a distribution-layer shift than a near-term earnings event. The economic value sits with whichever assistant becomes the default starting point for financial intent, because that controls engagement, not just execution; that is structurally favorable for GOOGL if Gemini can become a trusted high-frequency workflow, but the revenue conversion is still soft.

The more interesting second-order effect is competitive pressure on standalone broker UIs and research portals. If prompts become the front door, brokerage brands risk becoming back-end plumbing, which can compress app stickiness and raise customer-acquisition costs for consumer brokers while benefiting platforms that own authentication, custody, and APIs. FRBA is not a clean read-through; small banks get no obvious operating leverage here unless they can productize similar workflows.

The contrarian point is that AI can suggest ideas faster than humans, but this integration still leaves a human approval step, which limits turnover and keeps the monetization curve shallow. The study cited also flags sizing/risk control as the weak link, so the first public misfire could trigger heavier compliance overlays across the sector and slow adoption for months. Over 6-18 months, the winners are likely the wallets and rails, not the models.

Key Decisions for Investors

• No immediate position in FRBA; this is too indirect to underwrite as a balance-sheet or earnings catalyst. Treat as a watch item only if management later discloses AI-driven deposit or brokerage cross-sell.
• Modest tactical long GOOGL on pullbacks versus QQQ for a 6-12 month horizon: this is an option on Gemini becoming a high-intent consumer workflow, but size should stay small because direct monetization is not yet visible.
• Do not chase the brokerage/fintech angle yet; wait for adoption metrics from self-directed platforms. If AI-assisted prompts materially lift trade frequency without higher support costs, then revisit longs in IBKR or SCHW on stronger operating leverage.
• Set a regulatory alert on EU suitability/AI-guidance commentary. Any BaFin/ECB pushback would be the fastest falsifier for the adoption thesis and could pressure the whole AI-finance interface trade within days to weeks.