AI is the New Gatekeeper: Is Your Financial Firm Ready for the Age of Generative Optimization?
In a recent survey by Catalant consultant Ted Perkins, 60% of 500 consumers reported using AI in financial product decisions. When someone asks an AI assistant for a credit card recommendation or HELOC rate, it returns an answer instantly. Your firm is either in that answer or it isn’t.
This shift is already changing how financial services demand is formed. AI is no longer a research aid; it is increasingly acting as a gatekeeper, determining which products are even considered. This will meaningfully affect how financial services firms market themselves in a world where AI gatekeepers are already picking winners and losers.
Recently, Jeff Barden, Managing Director of Financial Services at Catalant, interviewed Ted, a consultant and former banker who helps organizations commercialize advanced technologies, including agentic commerce tools, to explore the implications of this shift and how firms can succeed in this changing landscape.
AI is the new ‘first filter’
For decades, financial institutions optimized for human behavior. Brand investment, SEO, and distribution strategy were built around how people search, compare, and decide.
That model is ending.
Consumers are already using tools like ChatGPT and Gemini to compare credit cards, evaluate banking products, and assess insurance options. AI systems return ranked responses based on how well available data matches the query. If a product does not surface, it effectively does not exist.
These systems evaluate offerings using training data and retrieved content. The process is opaque, varies across models, and is largely disconnected from traditional marketing signals. Brand strength and visibility in legacy channels do not guarantee inclusion.
The core problem for financial institutions is straightforward: most have no idea how they appear in AI-generated recommendations or whether they appear at all.
A new paradigm for digital visibility
Traditional digital visibility was designed for humans. AI changes that.
AI agents do not respond to brand campaigns, design, or reputation in the way people do. They perform a clinical evaluation, analyzing terms, structure, constraints, and product fit. This decouples brand marketing from product discovery.
Competing in this environment requires new practices:
- Generative engine optimization (GEO): Structuring product information and content so large language models can accurately interpret and recommend offerings. GEO is distinct from SEO, which targets human-facing search behavior.
- Headless legibility: Ensuring product data is accessible and parsable by AI agents, even when there is no human-facing interface involved. If information is not structured for machines to read, it is effectively invisible.
Listen: Ted Perkins evaluates the difference between human search and AI evaluation and how that leads to a focus on GEO.
AI recommendations are already diverging from legacy channels
The consequences of failing to adapt to GEO are not theoretical. As a real-world example demonstrates, AI models are already making editorial choices with profound financial implications.
When multiple leading AI models were asked the same question, “What are the best travel rewards credit cards?” the results varied significantly. Brands such as United, Hyatt, and Citi appeared inconsistently across models. Marriott Bonvoy, a top pick on traditional review sites like NerdWallet, did not appear in any of the responses.
This matters for revenue. Ted’s research found that 30% of consumers already use AI assistants to identify or compare credit card options, a figure that is likely to grow.
The takeaway is clear: visibility in traditional channels does not transfer to AI. High search rankings and prominent placement on review sites mean little if an AI model does not surface the product. The logic behind these recommendations is hidden, with no clear way to understand why one product appears and another does not.
Credit cards are just one example.
Beyond retail banking
This is not a consumer-only issue. The first-filter effect applies wherever decision-makers turn to AI:
- Wealth management: A prospective client asks an AI for a financial advisor in their area. The model returns a shortlist. Advisors not included are never evaluated.
- Commercial banking: An AI assistant evaluates treasury services based on available data, not relationships or brand recognition.
- Insurance: An independent agent queries an AI for a complex policy. The carriers surfaced depend on how clearly product details are structured for machine interpretation.
The implications extend across financial services. The distribution model itself is shifting.
Three no-regret moves
The AI landscape will continue to evolve, but waiting for certainty is not a strategy. There are practical steps financial institutions can take now that remain valuable regardless of how the market develops:
- Assess current visibility: Understand how your firm, brand, and products appear across major AI models today.
- Benchmark competitors: Identify where AI systems favor alternative offerings and why. These gaps are often structural, not substantive.
- Build the capability to adapt: A one-time audit is not enough. Firms need ongoing intelligence into how AI agents evaluate them and how that evaluation is changing.
Listen: Ted Perkins outlines the uncertainties in the AI landscape.
Is your business ready?
Generative AI is already shaping how financial decisions are made. The shift is subtle, but the consequences are material.
The question is straightforward: do you know how your firm appears to AI systems today and whether you appear at all?
Firms that understand and optimize for this new reality will stay relevant. Those that wait risk being filtered out entirely.
Learn more about how Catalant helps financial institutions navigate this shift.
Get in touchMost consumers now utilize AI assistants for financial decision-making, shifting the “first filter” of product discovery from human-centric search to machine-led evaluation. Research by Catalant consultant Ted Perkins indicates that traditional search engine optimization (SEO) targets human behaviors like brand recognition and visual design. In contrast, Generative Engine Optimization (GEO) focuses on structuring product data so large language models can clinically analyze terms and constraints. Financial firms must transition to GEO to remain visible, as products that do not surface in AI-generated recommendations effectively cease to exist for a growing segment of the market.
AI models perform a clinical evaluation of product data that is largely decoupled from legacy marketing signals and traditional brand strength. Top-ranked products on legacy sites like NerdWallet frequently fail to appear in ChatGPT or Gemini responses. This divergence occurs because AI agents analyze “headless legibility”—the machine-parsable structure of data—rather than human-facing reputations. Consequently, firms must ensure their technical documentation is accessible to AI scrapers to prevent being filtered out of shortlists for credit cards, wealth management, and insurance policies.
Financial services is shifting toward a model where AI agents, not human intermediaries, act as the primary gatekeepers for complex treasury and advisory services. If product information is not structured for machine interpretation, even established commercial banks risk invisibility during the “first filter” stage. Whether a prospective client is seeking a financial advisor or a complex insurance policy, the AI model returns a shortlist based on retrieved training data. This shift demands that firms provide granular, structured data to ensure AI assistants can accurately identify and recommend their service offerings.
Waiting for market certainty is a liability; therefore, firms must immediately audit their visibility across major large language models. Take three critical steps: assess current brand appearance in AI models, benchmark competitor visibility, and build an ongoing capability to adapt to changing AI evaluation logic. Because AI recommendations vary across models and change as training data updates, a one-time audit is insufficient. Financial institutions must develop continuous intelligence to understand why AI systems favor specific competitors and close those structural data gaps.
The gatekeeper effect extends beyond retail banking, as AI models now generate the shortlists used by high-net-worth clients and independent agents. Prospective clients often query AI for local financial advisors or carriers for complex policies. If an advisor or carrier is not included in the initial AI response, that professional is never even evaluated by the human decision-maker. To mitigate this risk, firms must ensure their value propositions and specialized expertise are documented in formats that AI models can ingest and rank during the clinical evaluation process.
