Retail
Questrade became the first Canadian brokerage to let customers connect their live account to Claude or ChatGPT directly via Model Context Protocol, letting an external AI assistant read real portfolio data and draft trades for in-app approval
CEO Edward Kholodenko unveiled the MCP integration alongside commission-free US options and pre-IPO access at a Toronto event, positioning it against Wealthsimple's prediction-market push and Robinhood's in-house agentic trading account as brokerages compete on how deeply AI gets to touch a real account.
The detail that separates this from the run of brokerage AI announcements this year is the protocol choice. Robinhood's Cortex and eToro's Tori are both AI agents built and hosted by the brokerage itself, and every other major launch this year has followed that pattern: the broker builds the model layer, the broker owns the interface, the AI is a feature of the app. Questrade instead built a Model Context Protocol connector, meaning a customer's own instance of Claude or ChatGPT, running outside Questrade's infrastructure entirely, can read that customer's actual portfolio holdings, watchlists, and market data and draft a trade for approval back inside the Questrade app.
That is a materially different trust boundary than a brokerage-hosted chatbot, and it is worth being precise about what does and does not change. The account data still only moves with explicit user authorization through MCP's permission model, and trades still require in-app approval before execution, so Questrade has not handed away custody or execution authority. What has changed is who controls the reasoning layer sitting between the data and the trade draft. A brokerage-hosted AI agent is bounded by whatever guardrails and data the brokerage chose to expose it to. An MCP connection to a general-purpose assistant the customer already uses for everything else in their life is a fundamentally more open-ended integration, for better and worse.
CEO Edward Kholodenko's framing, "we were fintech before there was fintech," and President Rob Galaski's "we're not looking to do what everybody else does" both read as a deliberate contrast with the Robinhood and eToro approach of building a proprietary in-house agent. Whether that is genuine product differentiation or a smaller player choosing the integration that costs less to build than training and hosting a dedicated trading model is the open question the announcement itself does not answer, and Questrade's incentive to frame a cheaper build as a bolder one is obvious enough to flag.
The regulatory backdrop gives this extra weight. Questrade is simultaneously pursuing prediction markets pending approval from the Canadian Investment Regulatory Organization, the same body that will eventually have a view on what it means for a third-party general-purpose AI model, not built or controlled by the regulated brokerage, to have read access to live client account data and the ability to stage trade orders. Wealthsimple is rolling out prediction markets this summer under the same regulator. CIRO has not yet had to write a rule for an MCP-connected brokerage account, and this launch means it now does.
Read the original: The Globe and Mail - Questrade became the first Canadian brokerage to let customers connect their live account to Claude or ChatGPT directly via Model Context Protocol, letting an external AI assistant read real portfolio data and draft trades for in-app approval. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.