Nearly eighteen months after its official announcement, Ava — Commerzbank’s AI-powered virtual assistant with a human-like avatar — has finally begun rolling out in the bank’s mobile app. Built on technology from Microsoft and its partner OpenAI, Ava is embodied as a German-speaking digital character modeled on a real actress to mimic natural gestures, facial expressions, and overall presence.

So what can this highly anticipated assistant actually do? The reality is somewhat underwhelming. Ava can answer natural-language questions about Commerzbank’s products and services — such as account types or credit cards — and can even provide simple comparisons between different options.

She also offers some basic personalization: users can ask about their financial situation or recent transactions. Ava can trigger a handful of actions, too, like ordering a credit card, locking or unlocking an existing card, or adjusting spending limits. But anything more “complex” is immediately handed off to a call center — along with the usual wait times and frustrations.

This raises the obvious question: why would consumers choose to interact with an avatar, no matter how polished, for such elementary tasks? And the irony is hard to miss: the moment a user needs help with something more sensitive — the kind of conversation where an animated, human-like interface might actually add value — they’re bumped over to a faceless phone operator.

Commerzbank’s official messaging offers a clear clue as to why ambitions were kept in check: security is the bank’s top priority. And the safest way to reduce risk and avoid errors, it seems, is to tightly limit what the AI is allowed to do. But once you “intellectually” hobble the system, what’s the point? Assigned real advisory responsibilities, Ava might have demonstrated her potential. Instead, as the end product of an 18-month effort, the gap between expectation and reality is hard to ignore.

Aware of the limits of traditional, passive personal finance tools, Québec-based Desjardins partnered with specialist Personetics to build a more proactive, AI-powered approach designed to deliver practical, personalized advice for every customer.

In a sign of just how confident the institution is in its strategy — and unlike previous, similar initiatives — Alvie, the new virtual assistant, appears to have been rolled out to all customers from day one, with no opt-out. She now occupies prime real estate on the mobile app’s home screen. At most, Desjardins allowed for a short transition period to help users get used to the upcoming change.

For the most part, the intelligent agent keeps a strong focus on budgeting — a continuation of earlier PFM features. Early improvements in this area anchor Alvie’s promise of financial guidance: helping users create financial plans with simple, user-friendly tools, plus detailed, contextual insights not only about their income and spending but also the habits that emerge from those patterns.

Even the system’s most original capabilities remain centered on budget management. Many of Alvie’s proactive recommendations — produced by analyzing transaction data to make them more relevant and valuable — focus on budget events. These range from alerts on overspending and specific optimization suggestions to positive reinforcement when users stick to their preset limits.

But Alvie also reacts to a broader range of everyday situations. She might prompt users to double-check an unusual transaction or consider switching providers after a service price increase. At launch, roughly twenty distinct message types — delivered in various formats — were programmed to respond to the common surprises of daily financial life, and that list will likely grow as Desjardins gathers more user feedback.

Crucially, the system lets customers rate the usefulness of each piece of advice. Those scores help tailor future interactions to individual preferences — and, at a broader level, guide the bank in refining the areas where expectations are highest.

Desjardins claims that customers have embraced Alvie enthusiastically, but it’s worth treating that optimism with caution. AXA Banque tried a similar concept in 2019, and despite a strong start, the project was eventually dropped, reportedly due to low usage after the initial burst of curiosity. AI may have improved the implementation this time around, but the real challenge remains: keeping customers engaged and preventing fatigue by making the assistant capable of tackling personal finance in all its dimensions.

Until now, AI’s early steps in personal finance have been cautious at best. With its brand-new assistant, the UK’s Starling Bank is beginning to sketch out what a true everyday financial companion could look like — one designed to genuinely help people manage their money.

Building on two earlier initiatives — one focused on analyzing financial flows and another on detecting fraud — the neobank is now aiming for a broader, more useful solution. It starts with a familiar foundation: a more advanced search engine that lets users query their transaction history.

On top of that, Starling adds another straightforward, if hardly groundbreaking, feature: a conversational interface for navigating the app’s settings and services in natural language. Users can ask practical questions like “What’s my card PIN?” or “How do I set up Apple Pay?” without digging through menus or dealing with a chatbot that understands only half of what’s asked.

A step up in sophistication comes with what Starling calls “agentic” capabilities. Extending the conversational logic further, the assistant can perform actions directly from free-form commands: “Set aside £500 and top it up monthly for a trip to Paris in July,” or “Create a budget of this amount for my everyday spending.” Nothing revolutionary — but it lowers the friction of tasks already available in the app.

Amid these fairly standard features, one clever touch stands out: users can ask the assistant to generate a quiz about their personal finances. It might ask which merchant they visited most in the past 30 days, or which spending category dominates their budget. It’s a simple, smart way to spark self-reflection — something traditional dashboards struggle to achieve without an engagement boost.

But what truly sets Starling’s assistant apart, at least in its current promise, is its focus on people’s real-world concerns rather than just banking functions. The framing centers on helping users save, manage their budgets, improve accessibility (especially for those with disabilities), and address vulnerabilities such as gambling addiction or debt problems.