Corporate and investment banks have invested heavily in strengthening their risk and capital frameworks to improve accuracy and transparency, particularly in the back book. But familiar gaps persist at the front end, notably limited incentives at origination to actively promote capital efficiency and limited regulatory and capital expertise.
Picture a more efficient origination process: A relationship manager initiates a deal in the front-office tool, and the system ingests the client’s financing need alongside key characteristics such as credit profile, sector, existing exposure, collateral availability, and strategic importance. An AI agent then generates multiple structuring options and transparently shows the impact of each on return on capital, capital consumption, and pricing, alongside critical arguments to prepare the client proposal.
Such a “capital-light” structure empowered by AI allows a relationship manager to engage the client with a set of optimized, data-backed alternatives rather than a single “standard” proposal. Structuring decisions that materially affect profitability and capital efficiency are embedded directly into the origination process—not just addressed later through slow escalation to the risk or finance functions.
According to McKinsey analysis, most advanced banks applying this approach can increase return on tangible equity by one to two percentage points, while capital-savvy relationship managers can typically originate products with up to 10 percent lower risk-weighted-asset (RWA) density than their peers. In fact, pilots suggest that agentic AI tools deployed at scale at the front line could reduce RWA density by an additional 5 percent across the board. We found that the combined 15 percent decrease in RWA density has been proven in underwriting pilots with European global systemically important banks (G-SIBs), where relationship managers were given AI “capital-light” tools for origination by comparing the selected final structure with the original client request received by the relationship managers. For banks facing sustained pressure on multiple fronts—from higher funding costs to tighter regulatory scrutiny and ever-increasing client expectations—agentic AI promises significantly greater value creation than incremental efficiency gains.
How AI can drive capital discipline at origination
Until recently, providing advanced risk and capital insight on individual transactions in near-real time was out of reach for most institutions. That constraint is now beginning to lift. Advances in generative and agentic AI enable a new operating model in which risk and capital expertise can be surfaced directly to the front line when decisions are made.
That means autonomous agents can not only reason across everything from client objectives to transaction structure, risk appetite, regulatory constraints, pricing, risk-adjusted return and capital impact, but continuously adapt as inputs evolve. Applied to origination, this makes it possible to embed current risk and capital logic directly into deal workflows, enabling the long-sought optimization of the new book in real time.
Capital-light structures are especially important as active regulatory developments (such as Basel IV implementation) add further complexity to the capital treatment for large corporates. The value of having the perspective of risk and finance available directly when origination decisions are being made, and specific to each individual client and need, is rising sharply.
A capital-light approach fundamentally shifts the management of capital efficiency from being a retrospective control function to being a live decision input. Rather than relying on sequential handoffs between the front line and risk, leading banks are embedding dynamic risk and capital guidance directly into existing platforms for customer relationship management and origination. The front line retains ownership of client outcomes, and risk perspectives are continuously updated, consistently applied, and transparently documented as decisions are being made.
Grasping the early-mover advantage
It’s tempting to compare the emergence of AI with another seismic shift to the banking sector: digital transformation. Yet while digital transformation also promised competitive advantage through speed, reach, and convenience, relatively slow consumer adoption—especially among older customers—gave banks time to adapt. This is not the case with AI. Leading banks are already starting to fundamentally change how the frontline interacts with risk and finance, with technology making it possible to move from episodic, escalation-based engagement to continuous, real-time collaboration at the point of decision.
While the AI transformation is still in its early days, institutions active in corporate and investment banking see it not as an operational improvement but as a strategic prerequisite for sustainable value creation. Relationship manager copilots are not only recommending alternative product structures to meet client financing needs but helping relationship managers with contractual clauses and integrating with pricing engines to consider the most accurate possible cost of capital. That offers clients products that optimize pricing as well as covering their financing needs.
A few fast-moving institutions are beginning to scale these approaches, accelerating execution and improving consistency without weakening risk discipline. That’s critical—because these advantages are unlikely to remain exclusive. Over time, the ability to embed real-time risk and capital perspectives into origination will become a baseline expectation, even though each leading bank will have its own differentiating capital-light initiatives. Furthermore, early movers can shape standards and build organizational muscle memory, while late movers could be constrained by slower processes and less-responsive interaction between the front line and risk.
Lorenzo Serino is a senior partner in McKinsey’s New York office, where Marco Vettori is a senior partner in McKinsey’s Milan office, where Enrico Muti is a partner in McKinsey’s Milan office, where Javier Martinez Arroyo is a partner in the Paris office, where Alejandro Gimeno is an associate partner in the Madrid office, and Vuk Vuksanović is a consultant.
Copyright © 2026 McKinsey & Company. All rights reserved.