rewrite this title The Consumer Credit Stack Is Being Rebuilt in Real Time – Finovate

With new enabling technologies like stablecoins and AI moving quickly and classic fintechs like Mint.com and Dwolla making their exits, it feels like fintech is entering a new era. This is especially true in lending, where new capabilities are enabling faster, more efficient, and in many cases more customer friendly tools than we had five years ago.

Looking back at the dawn of the decade, most lending innovation focused on digitizing the application process, facilitating the onboarding process, and turning loans faster. While some of those elements are still in place today, lending has changed with better intelligence, different distribution, and new infrastructure layers underneath credit itself.

Here’s a look at what’s changed:

Underwriting is becoming continuous instead of episodic

We used to think of the FICO score as the gold standard in underwriting. Today, however, underwriting is no longer done as a snapshot in time. Instead, lenders are using cash flow underwriting to get a view of the borrower’s creditworthiness over time by considering their account balance, overdraft occurrences, loan repayments, and other risk indicators.

Cash flow underwriting is becoming increasingly common, especially as consumers become more comfortable with open banking and the concept of sharing their financial data across platforms.

Embedded lending changed consumer expectations

Embedded lending itself is not new. Uber, for example, began experimenting with vehicle financing for drivers as early as 2014. What’s changed is how targeted, contextual, and embedded these lending experiences have become.

Today, financing is increasingly surfaced directly within the software platforms, marketplaces, and operational tools where consumers and businesses already spend their time. Point-of-sale platform Toast, for example, uses merchants’ daily sales data to underwrite loans and proactively surface financing offers within the Toast platform itself.

As consumers and businesses become more accustomed to contextual lending experiences like these and embedded buy now, pay later options they are relying less on traditional bank websites or standalone loan marketplaces to search for credit products.

The interface layer Is shifting

In addition to competition from software platforms and merchant ecosystems, a third distribution channel is beginning to emerge in lending: large language models (LLMs).

Consumers are increasingly turning to platforms like ChatGPT, Claude, and Gemini for both information and guidance and decision-making, including financial decisions. As these tools become more integrated into consumers’ daily lives, many borrowers may begin consulting an AI assistant before visiting a bank website or browsing a loan marketplace. Instead of searching manually for financing products, consumers may increasingly ask an LLM to help evaluate their situation and recommend the most suitable lending option.

That shift becomes even more significant as financial data aggregation moves into these environments. Through Plaid’s partnership with OpenAI, for example, ChatGPT can now aggregate and contextualize a consumer’s financial accounts, giving the platform a much richer understanding of cash flow, spending behavior, obligations, and financial goals.

As a result, the lender may still technically originate and hold the loan, but the customer relationship shifts to the interface layer. In this emerging model, the LLM becomes the discovery engine, recommendation layer, and engagement channel sitting between the consumer and the financial institution.

What scales vs. what doesn’t

Looking back at the lending technologies demoed on the Finovate stage five years ago, there is a noticeable divide between the ideas that generated excitement in the moment and the solutions that ultimately achieved scale.

Many of the products that struggled to move beyond the demo phase shared a common challenge: they required consumers to significantly alter their existing behaviors, communication methods, or digital environments. Metaverse-based banking and lending experiences, for example, were fun to watch on stage, but they never aligned with how most consumers wanted to interact with financial products in everyday life. In many cases, they required users to adopt entirely new platforms, devices, or behaviors before their value could even be realized.

By contrast, the lending solutions that have scaled most successfully are the ones that meet consumers where they already are. Buy now, pay later (BNPL) is perhaps the clearest example. Rather than requiring consumers to seek out financing separately, BNPL options are surfaced directly at checkout within the shopping experience itself. As a result, installment financing has become an expected feature for many higher-ticket purchases rather than a niche alternative payment method.

What credit looks like by 2030

Five years from now, much of today’s lending ecosystem will still look familiar. Regulated financial institutions will continue to originate loans, underwriting will remain central to managing risk, and compliance will remain a critical consideration not only for lenders, but also for fintech partners, platforms, and emerging distribution channels.

What may look very different, however, is the interface layer between the consumer and the lender.

Consumers may interact less directly with banks and more through AI assistants, software platforms, wallets, and embedded ecosystems that help evaluate financing options on their behalf. As LLMs become more integrated into everyday decision-making, they may fundamentally reshape how consumers discover, compare, and select credit products. In that environment, traditional loan marketplaces could become far less relevant as financing recommendations are surfaced contextually and conversationally through AI-driven interfaces rather than through manual product searches.

Photo by Silvio Pelegrin

Views: 41