SMBC Group and Sakana AI Launch Automated Strategic Proposal System to Scale Wholesale Banking Advisory
Sumitomo Mitsui Financial Group (SMBC Group) has deployed its "Automatic Proposal Generation Application" into production, a move that marks a vital step in automating the high-stakes world of corporate strategic advisory. The application is the first production-ready output of the strategic partnership established with Tokyo-based Sakana AI in May 2025.
The launch represents another step in the modernization of Japan’s "megabanks," which are increasingly looking to generative AI to protect margins and modernize labor-intensive business models. By integrating advanced AI agents into the core wholesale workflow, SMBC Group aims to standardize complex research and planning tasks, moving the bank closer to its goal of becoming a tech-driven global leader.
This technological leap addresses a critical bottleneck in the Japanese financial sector: the reliance on individual expertise to manage increasingly complex corporate management issues in an era of rapid digital transformation.
1. Modernizing the Financial Advisory Model
As Japanese corporate clients navigate an era of unprecedented volatility, the demand for sophisticated, data-driven financial advice has surged. Historically, providing this level of insight has been a "person-dependent" (属人的) endeavor, where the quality of a proposal was inextricably linked to the experience and research capacity of an individual banker. This model is no longer sustainable as digital transformation (DX) accelerates, creating a gap between the speed of client needs and the capacity of traditional human-centric workflows.
SMBC’s new system seeks to gain a competitive advantage over domestic and global peers by industrializing high-quality advisory services. This shift ensures that every client—regardless of the specific banker assigned—receives a standardized, top-tier strategic output.
By moving beyond these "person-dependent" structures, SMBC is positioning itself to deliver high-quality value to all customers with a speed and precision previously unattainable in wholesale banking.
2. Technical Architecture: Multi-Agent Synergy and Analysis
The "Automatic Proposal Generation Application" distinguishes itself from general-purpose AI by utilizing a sophisticated "multi-agent" architecture. This system mimics the iterative, collaborative process of a professional advisory team. The architecture involves multiple AI agents that interact with one another to refine proposal content in stages, effectively acting as a feedback loop between a "junior" researcher and a "senior" editor.
The process is synthesized through the following specialized agent roles:
- Information Gathering and Analysis: Agents autonomously aggregate vast arrays of financial and non-financial data, identifying trends that might escape human observation.
- Hypothesis and Story Building: These agents synthesize the raw data into logical, strategic narratives tailored to the client’s specific management challenges.
- Quality Evaluation and Fact-Checking: Crucial to the iterative process, these agents review the output for accuracy and professional rigor, refining the proposal through successive stages.
A defining feature of this system is its ability to offer "objective points of view" by analyzing disparate datasets. This often results in strategic recommendations that exceed customer expectations (お客さまの期待を超える), providing bankers with fresh perspectives to present to their clients. This technical efficiency refocuses human labor toward the highest-value aspects of the relationship.
3. The "Human-in-the-Loop" Strategy: Redefining the Banker's Role
The automation of information collection and document drafting is a strategic reallocation of SMBC’s human capital. By lifting the heavy administrative burden from the front-line staff, the bank is empowering its professionals to act as true consultants rather than data aggregators.
Bank employees are now expected to focus their expertise on Core Professional Focus Areas:
- Multi-faceted analysis: Layering human intuition and deep industry experience over the AI’s objective data findings.
- Formulation of essential problem-solving measures: Crafting highly nuanced, bespoke solutions that address the root causes of a client’s business dilemmas.
This "human-in-the-loop" approach is the engine behind SMBC’s ambition to be a "Trusted Partner" and to establish "Japan's No. 1 Corporate Business" by providing a level of service that combines machine speed with human judgment.
4. Ecosystem Integration and Future Roadmap
The deployment of the Proposal Generation AI marks the completion of an "end-to-end" digital workflow for SMBC’s wholesale business. It works in tandem with the "Sales Promotion AI Application" released on April 24, 2026. While the earlier application focuses on identifying and surfacing business leads, the new application provides the strategic depth required to close those deals. This synergy maximizes the total value of the wholesale operation, from initial lead generation to final strategy delivery.
Looking forward, SMBC Group intends to sequentially implement AI agent technology across other business domains, refining its operations through a growing library of case studies and functional improvements. Through these advanced implementations, SMBC Group reaffirms its commitment to sustainable growth and its vision of being a "Trusted Partner connecting the world from Japan."
Norbert Gehrke
