Could embedded financial optimization recreate Yu'e Bao at scale?

McKinsey's Ramji Sundararajan and Uzayr Jeenah, with Alana Ellis, point to Yu'e Bao as a real-world precedent for what agentic finance could do at scale. By 2017, they write, Yu'e Bao "had become the world's largest money-market fund, with about $268 billion in assets under management and more than 260 million users, many of whom had never actively invested before." McKinsey calls it "an early example of embedded, agent-like behavior: automating financial optimization at the system level" — even though, as the authors stress, "Yu'e Bao wasn't labeled AI."

The lesson McKinsey draws is structural, not technological: when financial optimization is embedded into a product consumers already use — rather than offered as a separate decision they must opt into — adoption can be extremely fast and friction can drop sharply, because the system, not the user's attention, is doing the optimizing. McKinsey calls Yu'e Bao "a cautionary blueprint for what agentic finance could enable — and what regulators may soon scrutinize," explicitly pairing the upside ("rapid adoption, reduced friction") with the downside ("new systemic risks") rather than treating embedded optimization as straightforwardly good.

This precedent gives empirical weight to the mechanism described in Will AI agents break the deposit and credit card inertia?: Yu'e Bao shows that embedded, inertia-breaking optimization has already moved hundreds of billions of dollars in assets once, without being branded as AI at all. It also complicates How willing are people to let AI control their finances?: Yu'e Bao's 260 million users suggest that discomfort with giving an agent-like system access to money can coexist with — or even be bypassed by — sufficiently embedded, low-friction product design.

The excerpt doesn't say what those "new systemic risks" were or are, nor whether Yu'e Bao's scale came from genuine agent-like autonomy or simply an attractive, well-marketed savings product; McKinsey's own framing leaves the causal mechanism — embedding versus marketing versus yield — unresolved, so the analogy to AI-driven agentic finance is suggestive rather than demonstrated.