AI banking goes real-time on compliance

Banks and fintechs like Slash are pragmatically integrating AI to streamline back-office operations and enhance customer experience, achieving significant cost reductions and operational efficiency. Victor Cardenas of Slash highlights their agentic AI approach to automate repetitive tasks, aiming for the highest EBITDA margins by lowering opex and enabling more competitive customer rates, while their AI Chief of Staff product leverages natural language to simplify user interactions, though most AI power remains behind the scenes. This pragmatic AI adoption is accelerating development cycles and automating mundane workflows despite regulatory and legacy system challenges, underscoring the necessity of modern, cloud-native platforms for effective integration.

Agentic AI is moving beyond basic conversational bots to automate complex, multi-step workflows in banking operations, including billing, underwriting, and financial crime investigations. Companies like Backbase, through its acquisition of Kasisto, and Paddle are embedding AI with governance and regulatory controls directly into their platforms, enabling intelligent resolution workflows that comply with stringent audit requirements. At First National Bank of Omaha, agentic AI has halved investigation times for fraud and AML alerts by automating evidence gathering while preserving human oversight, exemplifying how AI acts as a force multiplier rather than a replacement in regulated environments.

Effective AI integration in banking demands robust governance frameworks and operational resilience to manage regulatory complexity and scale across global platforms. Revolut’s approach emphasizes centralized AI governance through a single gateway that enforces consistent policies and cost controls across hundreds of teams, enabling seamless compliance and scalability for over 70 million customers. Their strategy includes fallback chains for generative AI models to maintain uninterrupted service and right-sizing models to optimize cost without sacrificing quality, reflecting the critical importance of detailed monitoring and control planes to prevent silent failures in AI-driven workflows.

The future of SMB banking is rapidly evolving from a focus on product proliferation to delivering deeply personalized experiences through strategic fintech partnerships. As Deepa Chatterjee of U.S. Bank explains, the bank now prefers collaboration over acquisition, embedding financial services that streamline small business operations and enhance digital capabilities. This shift is powered by AI-driven personalization, which processes vast contextual data to tailor financial advice and interfaces dynamically to individual SMB needs, moving beyond static models to truly adaptive financial management.

Embedded finance is emerging as a dominant revenue engine within B2B SaaS platforms, often eclipsing traditional subscription income. Companies like Toast and Shopify demonstrate this trend, with over 80% and 73% of their revenues respectively derived from embedded fintech solutions, including lending and payments. The market remains ripe for growth, with BCG and Adyen estimating a $185 billion addressable embedded finance opportunity, yet less than 20% is currently captured, highlighting vast untapped potential especially in embedded lending, which practitioners identify as the next frontier.

Success in embedded finance hinges on sophisticated integration strategies and robust risk management amid a fragmented software ecosystem. With over 20 major accounting platforms globally and top vendors controlling less than 55% market share, unified API aggregation layers become essential to scale cost-effectively, as direct integrations can cost up to $150,000 annually each. Moreover, sponsor bank selection and vigilant risk controls are critical, as evidenced by the 2024 enforcement cycle that saw major failures like Synapse’s $65 to $95 million customer shortfall, underscoring the high stakes of embedded finance continuity.

Embedded finance is fundamentally reshaping SMB and B2B banking by relocating financial services into the software platforms businesses use daily, such as ERP and procurement systems, effectively shifting customer relationships from banks to software providers. Banks are responding by layering intelligent software ecosystems atop their trusted core infrastructures—like Inlogik’s Spend Platform—to deliver seamless, integrated financial workflows. As Charles Crane notes, banks must leverage their inherent strengths in trust, capital, treasury expertise, and regulatory capability to remain relevant, evolving into connected financial experience providers embedded within business operations.

Citi exemplifies proactive regulatory engagement by collaborating with European regulators early in the rule-making process across 23 countries, allowing it to shape evolving standards and embed compliance into its strategic planning. This approach, combined with a vertically integrated global operating model, enables Citi to deliver standardized AI-enhanced solutions rapidly while tailoring them to local regulatory requirements, achieving a 30-40% faster deployment cycle through AI integration in development.

The Dubai International Financial Centre (DIFC) leads as the world’s first AI-based financial hub, pioneering the integration of AI into regulatory frameworks and market infrastructure. This controlled environment allows banks to pilot AI-driven services, rigorously test model accuracy and governance, and strategically navigate regulatory complexities before scaling regionally, underscoring the critical role of resilience and adaptability in meeting global AI-driven financial demands.

Backbase’s acquisition of Kasisto highlights a growing industry trend toward embedding agentic AI with built-in governance and regulatory controls directly into banking platforms. This integration addresses compliance challenges by enabling auditable decision logs, deterministic governance, and seamless core banking connectivity, which are essential for regulated finance and help reduce friction in AI adoption amid increasing regulatory scrutiny of automated decisioning.

Revolut’s AI governance model innovates by shifting from fragmented, model-specific controls to a centralized, use-case-based framework aligned with the EU AI Act, enabling consistent risk management across hundreds of teams and dozens of countries. Their governance layer enforces compliance through human-in-the-loop reviews, fallback mechanisms for generative AI failures, and strict data sovereignty controls, effectively managing operational risks and regulatory adherence for a global customer base exceeding 70 million.

Valley National Bank’s ambitious core banking modernization, which included migrating over 80% of its data center capacity to the cloud and establishing a centralized data hub, exemplifies how rebuilding legacy infrastructure is foundational to scalable AI integration and agile digital service delivery. This transformation not only enabled AI applications like the AML agent 'Tara' but also unlocked tangible business benefits—nearly $100 million in cost savings, $20 million in new recurring revenue, and over 100,000 employee hours—demonstrating that infrastructure modernization is a prerequisite for responsible and effective AI deployment rather than mere buzzword adoption.

Despite the clear imperative, many banks remain shackled by monolithic mainframe systems and layered patchwork technologies, a legacy of cautious avoidance of core overhauls due to cost, security, and scalability fears. This has left institutions struggling to meet modern customer demands for real-time payments and AI-driven capabilities, as these legacy stacks are neither agile nor fit for purpose. Industry experts emphasize that incremental updates—such as focusing solely on payments or onboarding—fall short of enabling meaningful innovation, underscoring the necessity of comprehensive end-to-end platform reconstruction or cloud migration to remain competitive.

Modernizing core banking systems is a complex, multifaceted endeavor that often multiplies project scope due to the deep entanglement of legacy platforms with numerous hidden processes. As one interviewee noted, replacing a core platform can create '50 new projects' for a CEO, reflecting the intricate dependencies within the bank’s operations. To overcome this, firms like Constantinople advocate for a holistic end-to-end platform rebuild that simplifies modernization, accelerates product time-to-market, and better enables AI adoption—transforming the core technology and operations layer independently from customer-facing and licensing components to facilitate global scalability without regulatory bottlenecks.