I have spent the better part of my career at the intersection of technology and financial services — first as an engineer building payment systems, then as a founder building fintech products, and now as an investor helping the next generation of fintech founders navigate one of the most complex and rewarding markets in technology. In all of that time, I have never seen a moment quite like this one.
The convergence of artificial intelligence and financial services is not a future possibility or a trend to watch. It is happening right now, at scale, across every segment of the financial services stack, and it is generating real business outcomes for early movers. At Swarm Capital, fintech is one of our three core investment pillars, and we have spent this year deploying significant capital into what we believe are the most important seed-stage fintech companies of the decade.
This piece is my attempt to share the analytical framework we use to evaluate AI-fintech opportunities, the specific segments where we see the greatest venture potential, and the criteria that separate the seed-stage fintech companies that will become category leaders from those that will struggle to achieve meaningful scale.
Why Financial Services Is the Perfect Domain for AI
Financial services has three characteristics that make it an unusually rich domain for AI application. First, it is deeply data-intensive. Every transaction, account action, market movement, and customer interaction generates data. Financial services firms are among the largest data generators in the global economy, and the value of that data — for underwriting decisions, fraud detection, personalization, and market making — is enormous.
Second, financial services is process-intensive in ways that map naturally to AI automation. Loan underwriting follows a logic chain. Compliance monitoring follows rule sets and pattern matching. Customer onboarding involves document processing and verification. These are exactly the kinds of structured, repeatable processes where AI systems can dramatically outperform humans in speed, consistency, and cost-efficiency.
Third, financial services is regulation-intensive, which creates both a barrier and an opportunity. The barrier: navigating financial regulation is expensive and complex, and AI systems deployed in regulated financial contexts must meet higher standards of auditability, fairness, and reliability than most other application domains. The opportunity: companies that build AI systems that are demonstrably compliant, fair, and auditable are building a moat that less sophisticated competitors cannot easily cross.
The Five Segments We Are Most Excited About
1. AI-Native Underwriting
Traditional credit underwriting is broken in a specific and important way: it relies on backward-looking data (credit history, income history, asset history) to make forward-looking predictions (will this person repay this loan?). AI-native underwriting uses a much richer set of signals — behavioral data, transaction patterns, real-time revenue data for businesses, social and professional network signals, and more — to build more accurate predictive models for creditworthiness.
The result is not just incrementally better underwriting — it is fundamentally different underwriting. Companies that have been historically underserved by traditional credit models (immigrants with thin credit files, small businesses with variable revenue, gig workers with non-traditional income streams) suddenly become creditworthy customers. The market expansion effect is enormous: AI-native underwriters are not just stealing share from traditional lenders; they are serving populations that traditional lenders have written off entirely.
Swarm Capital has invested in two companies in this category. Our portfolio company Pebble Credit exemplifies the model: AI-underwritten revenue-based financing for digital commerce businesses, using real-time platform data to extend credit to businesses that traditional bank credit teams cannot evaluate quickly or accurately enough. The default rates are competitive with traditional models despite serving businesses with no credit history, because the behavioral and revenue signals are more predictive than the backward-looking data that traditional models rely on.
2. Embedded Finance Infrastructure
The embedded finance market — financial services products offered by non-financial companies within their existing customer experiences — is estimated to exceed $7 trillion in transaction volume by 2030. Every software platform, marketplace, and digital business with a loyal customer base has the potential to offer financial products that deepen customer relationships and generate new revenue streams.
The bottleneck is infrastructure. Building embedded financial products from scratch requires banking licenses or bank partnerships, compliance infrastructure, fraud and risk management systems, and core banking ledger capabilities — none of which is the core competency of a retail software company or marketplace. This bottleneck is the opportunity.
Companies building modular, API-first embedded finance infrastructure are in the early stages of a massive market expansion. The analogy to Stripe's role in enabling internet commerce is instructive: just as Stripe eliminated the complexity of accepting payments online, embedded finance infrastructure companies are eliminating the complexity of offering banking, lending, and card products within any software application. Our portfolio company FlowFinance is positioned at exactly this intersection, and the growth we have observed since our seed investment in January 2025 has exceeded every projection.
3. Regulatory Technology and Compliance Automation
Financial services compliance is a $270 billion annual market globally, and it is primarily still driven by manual processes: armies of compliance officers, legal teams, and consultants doing work that involves substantial pattern matching, document review, and rule application. AI is beginning to automate this work in ways that generate 10x efficiency improvements and, perhaps more importantly, dramatically reduce the risk of human error in compliance-critical decisions.
The RegTech opportunity is particularly interesting because of the regulatory acceleration happening globally. The EU's AI Act, evolving US financial services regulation, and the emergence of novel asset classes that require new regulatory frameworks are all generating compliance complexity at a rate that exceeds the capacity of traditional compliance functions to absorb. Companies building AI systems that monitor regulatory changes in real time, map them to internal policies, and generate audit-ready documentation are in an enviable position.
Our portfolio company ComplianceOS serves mid-market and enterprise financial services firms across 60+ jurisdictions and has demonstrated that AI-native compliance automation can maintain zero regulatory findings for its customers while reducing compliance operational costs by 70%. The product is deeply embedded in customer workflows and generates strong net dollar retention as customers expand their usage across additional regulatory domains.
4. Cross-Border Payments and FX Infrastructure
Global cross-border payments represent a $150 trillion annual market, and the infrastructure underlying most of those payments is astonishingly antiquated. SWIFT, correspondent banking, and the FX interbank market were designed decades ago for a world where international payments were relatively rare and time-sensitivity was measured in days. For the digital economy — where global commerce happens at internet speed and SMBs operate across dozens of currency corridors simultaneously — this infrastructure is inadequate.
AI is enabling a new generation of cross-border payment infrastructure that routes transactions dynamically through the most efficient corridors, predicts FX rate movements to optimize settlement timing, applies AI-based AML and fraud detection that is both more accurate and more efficient than rule-based approaches, and handles the regulatory complexity of multi-jurisdiction payments through intelligent automation. Our portfolio company Meridian Pay is building exactly this infrastructure, and the growth trajectory since its seed round closure in February 2025 validates the market need emphatically.
5. AI-Native Wealth Management Infrastructure
The wealth management industry is experiencing a structural transformation driven by the democratization of previously institutional-grade investment approaches. Tax optimization strategies, alternative asset access, factor-based portfolio construction, and sophisticated rebalancing algorithms that were once the exclusive domain of multi-billion dollar family offices and institutional investors are now accessible at the individual investor level through AI-enabled platforms.
The opportunity is not just in direct-to-consumer wealth management — which has been a difficult market for pure-play robo-advisors. The more interesting opportunity is in the infrastructure layer: enabling registered investment advisors, broker-dealers, and independent financial advisors to offer their clients the same institutional-grade capabilities through an API-first platform that integrates with existing custodian and reporting workflows. This is the RIA enablement play, and it is a large, defensible market with strong network effects.
What We Look For in AI-Fintech Startups
Having invested in six fintech companies from the seed stage, we have developed a set of criteria specific to AI-fintech that we apply to every opportunity we evaluate.
Regulatory strategy from day one. Fintech companies that treat regulation as an afterthought invariably face expensive and time-consuming compliance remediation later. The best AI-fintech founders have thought deeply about their regulatory footprint before they have a product, have either hired compliance expertise or established relationships with specialized fintech legal counsel, and have designed their product architecture with auditability and compliance in mind. We will not invest in a fintech company where the regulatory strategy appears to be "figure it out later."
Banking or financial institution relationships. Many of the most important fintech products require partnerships with regulated financial institutions — banks, credit unions, transfer agents, broker-dealers — that provide the licensed infrastructure on which fintech products operate. These partnerships take time to establish and carry significant switching costs for both parties. AI-fintech startups that have already secured a committed banking partner, or that have founders with the relationships to establish one quickly, have a structural advantage over those that are starting from scratch.
Unit economics that survive scale. The worst fintech unit economics are those that look attractive at small scale but deteriorate as the business grows. Fraud losses, credit losses, compliance costs, and technology infrastructure costs all behave differently at scale than at the early stage. We build detailed bottoms-up models of unit economics under scale scenarios for every fintech investment, and we decline investments where the path to strong economics at scale is unclear or requires assumptions that seem heroic.
Proprietary data as a moat. The most defensible AI-fintech companies are those that accumulate proprietary data with each transaction or customer interaction — data that improves their models and that competitors cannot easily replicate. An AI credit underwriter that has processed 100,000 loans has better training data for its risk models than a new entrant starting from scratch. This data flywheel is the primary source of durable competitive advantage in AI-fintech, and we look for early evidence of it in every evaluation.
The Risk Landscape
No honest analysis of AI-fintech investment opportunities is complete without acknowledging the risks. Financial services is a regulated industry, and regulatory risk is real: a change in how regulators view AI-based underwriting, or new restrictions on the use of alternative data, could significantly affect the business models of companies in our portfolio. We actively monitor the regulatory landscape and work with our portfolio companies to build compliance programs that are robust to regulatory evolution.
Credit risk is a material risk for fintech companies in the lending and payments space. Economic downturns stress credit portfolios in ways that are difficult to model at the seed stage, and fintech companies that have not built conservative risk management practices can face existential losses during credit cycles. We require our fintech portfolio companies to stress-test their credit models against historical recession scenarios and to maintain conservative credit risk appetites until their models have been validated through multiple economic cycles.
Competitive risk from incumbent financial institutions is real but often overestimated by early-stage investors. Large banks are investing heavily in AI and have advantages in customer relationships, data, and regulatory infrastructure. However, the organizational and cultural barriers to innovation at large financial institutions mean that AI-native startups can move significantly faster, iterate more rapidly, and build products that are better suited to the specific needs of underserved customer segments. The key is to pick battlefields where the incumbents are structurally disadvantaged — and there are many.
The AI-fintech opportunity is generational. The convergence of more powerful AI capabilities, declining technology costs, and the structural need for financial services modernization is creating a window for seed-stage companies to build the financial infrastructure of the next decade. At Swarm Capital, this conviction guides some of our highest-conviction investment decisions, and we believe the returns from the companies we are backing today will validate that conviction over the next five to seven years.
