Citi AI Summit 2026: From productivity to democratization -- AI’s next phase in financial services

This year's Citi AI Summit discussion on the impact of AI on financial services kicked off with a key question: what are AI-native startups building that large language models (LLMs) cannot easily replicate?

To answer this and many other timely questions, I was joined by Gabe Stengel, CEO, Rogo; Rafael Loureiro, CEO, wealth.com; Patrick Vergara, COO, Norm AI; and George Sivulka, CEO, Hebbia.

Answers to the above and many other topics revealed a nuanced and evolving landscape as AI continues to impact financial services. Today’s moats are not brand scale or balance sheet strength, but rather last mile specificity, embedded legal and compliance capability and deterministic precision in highly regulated workflows. These advantages, however, were not framed as permanent. Panelists broadly agreed that as foundation models continue to improve, many of today’s perceived moats will erode.

The deeper, more durable opportunity lies elsewhere: using AI to fundamentally democratize access to high finance services that have historically been limited to the ultra wealthy and the largest institutions.

Rather than viewing AI as a productivity layer on top of existing financial services, the discussion positioned it as a redistributive force, one capable of changing who gets access to expertise, at what cost and with what level of quality.

Democratization as the defining
opportunity

One of the clearest points of consensus was that AI’s most transformative impact in financial services will be democratization. Today, the highest quality estate planning, tax optimization and financial advice is economically viable only for individuals with millions in net worth. This is not because the advice is inherently complex, but because it depends on scarce expert labor and manual processes.

AI changes that equation. By encoding expert reasoning into scalable systems, firms can deliver institutional grade financial expertise to individuals, small businesses and advisors who were previously priced out. This represents more than market expansion, it is a structural shift in how financial knowledge is distributed. As AI systems take on increasing responsibility for analysis, synthesis and planning, the baseline quality of financial services rises for everyone.

Temporary moats?

Panelists emphasized that building successful AI native financial products requires much more than general intelligence. It demands deep encoding of firm specific workflows, jurisdiction specific regulations and multi step compliance processes that may span dozens of actions and approvals.

These systems frequently must handle highly structured documents, interpret nuanced regulatory language and make decisions that mirror professional judgment. That depth of integration separates AI native firms from general platforms and horizontal tools. However, the panel was explicit that this moat is temporary. As foundation models become more capable and better at reasoning across complex contexts, last mile workflows that seem defensible today will become easier to replicate. The critical question is whether companies are building ahead of that curve or merely exploiting a short lived advantage.

Compliance as a source of revenue, not friction

Traditionally, compliance in financial services has been treated as a cost center but the panel argued that AI enables a fundamentally different approach. When compliance is embedded directly into AI systems and workflows, it becomes a revenue enabling asset.

AI native compliance infrastructure allows firms to take on additional business, move faster and operate in areas that were previously uneconomical due to regulatory overhead. Instead of asking how to minimize compliance burdens, leading companies are now asking how compliance can unlock new sources of growth. This shift reframes regulation as a strategic differentiator for companies capable of operationalizing it at scale.

ROI, economics and the end of
tokenmaxxing

On the economic front, the panel was united in declaring the end of tokenmaxxing. The early phase of AI adoption rewarded experimentation and tolerated massive spending on compute without clear business justification. That era is over. Enterprise buyers, regulators and investors now demand explicit mapping between AI spend and financial outcomes.

Vertical AI companies, especially in financial services, now carry the burden of proof. They must demonstrate not only that their systems work, but that they deliver measurable return on investment and could not have been built more cheaply in house. This has profound implications for product design, go to market strategy and pricing. Transparency about objectives, risks and measurable improvements is no longer optional.

Redefining the future of financial
services

Taken together, the panel painted a picture of an industry on the cusp of structural change. Incumbents retain enormous advantages, but they are not immune to disruption. AI native firms are rewriting the cost structure, accessibility and delivery of financial services, while forcing a reconsideration of compliance, security and pricing along the way.

The winners will not be those who simply add AI on top of existing processes, but those who rebuild workflows, economics and trust models from the ground up.

Are you a founder building enterprise-grade AI solutions? I’d love to talk! Reach out to me at jelena.zec@citi.com.