On August 25, 2026, Scalable Capital said it had opened its brokerage platform to major AI assistants, including ChatGPT, Claude, and Grok, under a new model it calls Agentic Investing. The company claims it is the first bank in Europe to do so, and it is letting clients connect an assistant, issue natural-language prompts, and then approve trades before execution (Scalable Capital newsroom).
What Scalable Capital Agentic Investing enables
Scalable Capital frames the launch as a full-featured layer on day one. According to the company, users can trade, set up savings plans, manage watchlists, and create price alerts, all by typing or speaking plain-language requests. The service also exposes a native search across stocks, ETFs, and derivatives, plus free access to news, real-time quotes, and historical price data. Deeper analytics flow in from Scalable Insights, including diversification checks, scenario analyses, sector and region breakdowns, and risk assessments (Scalable Capital newsroom, August 25, 2026).
The integration is assistant-agnostic. Scalable says clients can pick ChatGPT from OpenAI, Claude from Anthropic, or Grok from SpaceXAI, connect the assistant, and manage permissions inside profile settings. A key control remains: the user must approve any trade or savings plan before it is placed. Every action is visible in the app or on the web, which provides an audit trail that should help resolve errors and support compliance reviews.
In practical terms, the experience shifts from tapping through tabs to asking a question. A client could request a summary of energy exposure in their portfolio, then ask the assistant to adjust a savings plan if exposure drifts beyond a threshold. The approval gate stands between suggestion and execution, which is the right design for a tool that can still misread a prompt or miss context.
Why this agentic investing model matters
The strategic bet is clear: people will spend more time inside assistants than inside single-purpose apps. If that’s true, then a broker that treats the assistant as the primary interface wins mindshare without forcing users to relearn a UI. Scalable Capital’s move pulls the center of gravity toward the assistant and away from the app shell.
There is a historical rhyme here. In payments and account data, Europe’s PSD2 opened access for third parties via APIs, which catalyzed a wave of aggregators and new interfaces. This step does something different: it opens the front door to a general-purpose assistant that sits above many services. That could compress the value of bespoke app navigation and push brokers to compete on execution quality, insights, and guardrails rather than on UI flourishes. For readers unfamiliar with the earlier shift, the European Commission’s page on PSD2 outlines how regulated access reshaped banking from January 2018 onward (European Commission).
In that light, Scalable Capital Agentic Investing is less a feature and more a distribution change. If clients can ask one assistant to rebalance a portfolio, pay a bill, and schedule a cab, then loyalty may hinge on which services plug into that daily flow with the least friction.
Compliance and risk: from MiFID II to the EU AI Act
Two sets of rules loom over any AI-mediated trading flow in Europe. The first is investor protection under MiFID II. ESMA’s guidance clarifies when a firm is giving advice and must run suitability checks versus when it is executing client orders without advice. If an AI conversation crosses into advice, the broker’s obligations rise, including documenting why a recommendation suits the client’s profile (ESMA guidance, September 23, 2022).
Scalable’s design choice—client approval before any order is placed—helps draw a bright line and keep the human in the loop. It also provides a clear record of intent, which matters if an assistant misunderstands a prompt or hallucinates an action. The second rulebook is the EU AI Act, formally adopted by the European Parliament on March 13, 2024. The Act imposes transparency, logging, and risk management duties, with tougher requirements for high-risk systems. Finance features across several supervisory frameworks, so firms rolling out assistant integrations will be mapping obligations now and phasing controls as the Act’s timelines kick in (European Parliament).
For investors, the near-term risks are more basic. Misheard symbols. Fat-fingered quantities. Ambiguous prompts. The mitigation stack is familiar: permission scopes, read-only defaults, explicit confirmations, and easy revocation of access. Scalable Capital says all activities can be monitored in-app or on the web, which should lower the cost of catching mistakes. The next test is how the company handles edge cases: partial fills, corporate actions, and messaging when a requested instrument has low liquidity.
What Scalable Capital Agentic Investing changes for competitors
The launch pressures European brokers and banks to pick a lane. Do they integrate general-purpose assistants and push approvals through their own risk engines, as Scalable does, or do they keep assistants at arm’s length and build their own constrained chat layers? Assistant-agnostic support makes sense for reach, but it also means more permissions to manage and more prompt security to audit.
There’s also a data question. If most retail research moves into assistants, then content and analytics need to be assistant-readable. Scalable’s offer of free news, real-time quotes, and portfolio analytics inside the assistant context is a direct answer to that shift. The next step could be structured outputs, so a user can ask, “What changed since my last rebalance?” and get a crisp diff that translates into a queued order with one tap to approve.
Scalable Capital Agentic Investing also creates a standard for transparency that others will need to match. Clear logs, reversible permissions, and explainable prompts will not just be nice-to-have. They will be table stakes if clients are expected to trust an assistant with account access.
What investors should watch as agentic investing rolls out
Early adopters can use this launch without handing over the keys. The approval gate means you keep final say. Still, there are a few practical checks worth doing before connecting any assistant:
- Review permission scopes. Start with read-only access, then add trading rights if the experience proves reliable.
- Test ambiguous prompts. Try tickers with lookalike symbols and verify the assistant confirms the right instrument.
- Check audit trails. Make sure you can see every assistant-initiated action and revoke access with one click.
- Set small limits. Cap order size and savings-plan changes until you trust the flow under stress.
On the regulatory front, expect supervisors to ask for evidence that client approvals are informed and that AI interactions do not drift into unmonitored advice. Firms will also need to show how they detect and contain prompt injection or instruction conflicts across multiple assistants. The MiFID II suitability line and the AI Act’s logging duties will shape those controls. Readers wanting the legislative backdrop can revisit the AI Act’s adoption details and staged enforcement on the Parliament’s site, and the suitability guidance under MiFID II at ESMA.
One more parallel is useful. PSD2 turned banks into platforms by forcing data and payments access. Assistant access, by contrast, is elective. That makes adoption a signal. If client opt-ins grow and engagement rises, others will copy. If users stall at setup or bump into errors, brokers will retreat to safer, in-house chat tools.
Scalable Capital’s own framing is bold. Co-founder and Co-CEO Erik Podzuweit called Agentic Investing “the greatest technological shift in financial technology since internet banking” in the company’s announcement. That’s a high bar. The proof will be in day-30 retention, error rates on approvals, and whether clients keep asking an assistant to invest after the first novelty fades.
For now, the company has planted a marker. Scalable Capital Agentic Investing puts general-purpose assistants on the trading desk, with the human still confirming the final click. If that pattern holds, the brokerage app will become a control room, and the assistant will become the way most people invest. For more on this, see reuters.com and bloomberg.com and nytimes.com.