Wealthsimple's Penny: the AI assistant that won't trade without you

TL;DR: Wealthsimple unveiled Penny, a new AI assistant, at its "Wealthsimple Presents" event in Toronto on 5 November 2026. Penny can draft e-transfers, pay bills, flag duplicate credit card charges, suggest RRSP top-up schedules, and prepare buy or sell orders for equities and ETFs — but every action that moves money or executes a trade needs explicit confirmation from the customer first. Client data stays inside Wealthsimple's own systems rather than being shared with an outside AI provider, and by early 2027 customers will be able to connect their accounts to external tools like ChatGPT and Claude with controlled access levels. The real lesson isn't about Penny specifically — it's about a design choice every "AI that handles your money" feature has to make: how much does it do on its own, and where exactly does a human have to step in?

What Penny actually does

Penny sits inside the Wealthsimple app and handles tasks that would otherwise mean several manual steps: sending an e-transfer, paying a bill, spotting a duplicate charge across linked accounts, or mapping out a plan such as a recurring RRSP contribution schedule. For investors, it can also prepare an order to buy or sell a stock, ETF or other security. According to Wealthsimple's own team, the assistant uses a secure data feed connected directly to the company's systems, rather than screen-scraping account information the way some third-party finance apps do, and every action it takes or proposes is logged in an audit trail.

The feature everyone asking "is this safe?" should notice first is the confirmation step. Penny can research, draft and queue up an action, but a Wealthsimple product director was direct about where the line sits: hitting send on a transfer or a trade is "an action that the user alone can take." That's a meaningfully different default from an assistant that executes on your behalf unless you opt out — it's one that drafts on your behalf and waits for you to opt in, every single time.

Why banks keep building this feature differently

Penny is only the latest example of a pattern spreading across fintech: an AI layer bolted onto software that already moves real customers' money, rather than a model built from scratch. Wealthsimple launched Penny the same week several of Canada's big banks moved to block third-party AI agents from accessing customer accounts, citing security risk — while Wealthsimple went the other way and built its own first-party assistant it controls end to end. That gap between "block outside AI entirely" and "build our own AI, carefully" is exactly the kind of decision that needs YuSMP Group's generative AI integration expertise to get right: wiring a conversational assistant into an app that already handles live payments and trades is a very different job from adding a chatbot to a marketing site, because every mistake has a real account balance attached to it.

Wealthsimple's trading rival Robinhood took a looser approach with its own AI agents, which customers can switch to trade inside a dedicated account without asking permission for each trade — a design we broke down in Robinhood's AI agents: a risk lesson. Neither approach is simply "right" or "wrong"; they're two different answers to the same design question, and comparing them is a good way to see how much that one decision changes a product's risk profile.

The "secure by design" claims, and what to actually check

Wealthsimple has made a few specific claims worth separating out rather than taking as one blob of reassurance: client data stays within Wealthsimple's platform rather than being handed to an outside AI provider for processing; the assistant requires opt-in before a customer can use it at all; and every action it takes is recorded in an audit trail a customer or the company can review later. Those are concrete, checkable design choices, not marketing language — which is exactly the kind of detail worth looking for in any "AI-powered" finance feature you come across, instead of taking the word "secure" at face value.

It's also worth noting what's still ahead rather than already live: by early 2027, Wealthsimple plans to let customers connect their accounts to external AI assistants like ChatGPT and Claude, with what the company describes as granular access controls. That's a bigger trust decision than using Penny inside Wealthsimple's own walls, because it means a third party's AI model gets some level of access to real account data — a good one to watch once it actually ships, rather than assume works a certain way today.

How to size up any "AI that handles your money" feature

1. Find the confirmation step. Does the tool act automatically, or does it wait for you to approve each action? That one setting changes how much risk you're carrying by default.
2. Check where your data goes. Does it stay inside the app's own systems, or does it get shared with an outside AI provider? Penny's design keeps it in-house for now.
3. Look for an audit trail. A feature that logs what it did and why is easier to trust — and to challenge if something goes wrong — than one that just acts.
4. Separate the headline from the mechanics. "AI assistant launches" is the headline; "requires confirmation before every trade" is the detail that actually tells you how much control you're keeping.

The takeaway

Penny's launch is a small but clear case study in a question that's only going to come up more often as investing apps add AI features: automation and control aren't opposites, they're a dial, and every company building one of these tools has to decide where to set it. Wealthsimple's choice — draft freely, but never send without you — isn't the only possible answer, but it's a useful one to measure the next AI finance feature you see against.

This article is educational and is not financial advice, and nothing here is a recommendation to use or avoid any specific product. Details on Penny's launch and features are as reported by BNN Bloomberg and The Globe and Mail.