Financial institutions are moving rapidly from experimenting with artificial intelligence (AI) to applying it within real business workflows. In commercial lending, they are exploring how AI can help lenders find information, make faster decisions, and complete work more efficiently.
The next challenge is not simply selecting a powerful AI model. It’s about connecting AI securely to trusted information, business rules, and controlled actions across a complex financial services environment. The Model Context Protocol (MCP) has emerged as a significant option to address that challenge.
From experimentation toward adoption
AI adoption in banking is no longer limited to a small group of technology savvy players. According to the 2025 EY-Parthenon Generative AI in Banking survey, 77% of banks had launched or soft-launched generative AI applications, and nearly half had fully implemented at least one use case.
Momentum is also evident among regional and community financial institutions. Bank Director’s 2025 survey found that 62% of US banks with less than $100 billion in assets were actively experimenting with AI.
Commercial lending is an important part of this digital transformation. McKinsey’s 2025 research found that 52% of the surveyed institutions had prioritized generative AI adoption within their credit business. Use cases included early-warning systems, credit decisioning, credit-memo drafting, and customer engagement.
As adoption accelerates, the challenge is no longer whether to use AI, but how to use it effectively. This requires securely connecting AI solutions to the information and capabilities required to make that work efficient, secure, and reliable.
What is Model Context Protocol (MCP) and why does it matter for banking?
MCP was introduced by Anthropic as an open-source standard to address a growing problem: AI applications were becoming more capable but remained separated from the data and systems where useful context resides. The protocol provides a common way for AI applications to connect with external data sources, tools, and workflows.
Through MCP, AI applications can securely access approved enterprise data and services to support lending workflows, including:
- Retrieving and organizing information about a borrower, loan or portfolio
- Conducting due diligence searches for borrowers
- Analyzing and summarizing loan documents
- Applying organizational policies and business rules
- Preparing a recommendation for action or approval
MCP does not replace a financial institution’s existing lending systems, data platforms, or security controls. Instead, it can provide a standardized AI-facing interface for selected capabilities and expands the surfaces where they can be leveraged.
Financial-services workflows are distributed across many systems. A lender may need information from a loan-origination system, customer relationship management platform, core banking system, document repository, public-record source, credit platform, compliance system, filing service, or portfolio-monitoring application.
Imagine a commercial lending professional asking: “Review this borrower’s existing liens, identify potential conflicts, and recommend the next due-diligence step.”
Answering that question may require the AI-driven workflow to:
- Resolve the borrower against entity records
- Identify relevant jurisdictions
- Search UCC and other records
- Compare the findings with lender-defined policies
- Summarize the results with source context
- Highlight exceptions, uncertainty, or missing information
- Recommend next steps for approval
This is the practical promise of MCP: helping an AI application coordinate information and capabilities around the user’s objective rather than just answering questions. The same approach could help servicing or compliance teams review changes affecting an existing loan or lien perfection.
Leading loan origination system (LOS) providers, credit and risk platforms, document automation vendors, and lien management providers could expose selected capabilities through MCP to support the emerging AI-driven workflows.
Moving beyond application programming interfaces (APIs) to intelligent automation
Traditional integrations are important, but each new workflow can require additional development, data mapping, testing, and security review. MCP offers a common way to make approved capabilities quickly available to multiple AI-driven workflows.
The use of MCP is not to convert every API endpoint into an MCP tool. A more important question to ask is: What should the AI-driven workflow help the user accomplish?
An API for search may return records. A redesigned AI-based workflow could resolve a borrower, determine the relevant jurisdictions, perform the appropriate searches, apply customer-defined criteria, summarize the results, identify exceptions, and prepare the next action for review.
That is a fundamentally different experience from simply exposing an endpoint through a new interface. MCP can help financial institutions move from system integration toward workflow integration with potential to achieve greater efficiency.
From a security perspective, MCP connections, like APIs, can be configured to operate within established authentication, authorization, role-based access, data-protection, and audit controls.
AI governance and risk management considerations for financial institutions
MCP can make capabilities easier to access, but it does not make the underlying information reliable automatically. The quality of an AI-driven workflow depends on:
- Trusted and authoritative data sources
- Data currency and provenance
- Strong authentication and authorization models
- Auditability and traceability
- Human approval for consequential actions
- Monitoring, testing, and error handling
- Model risk and regulatory requirements
These considerations are especially important in lending. A useful AI-driven workflow must be intelligent, explainable, traceable, and grounded in trusted information.
The opportunity ahead
Financial institutions are entering a new phase of digital transformation. The next stage will require more than better models. It will require secure and governed connections between AI applications, trusted data, business rules, and enterprise capabilities.
MCP offers one potential foundation for that future. Its greatest promise is not simply faster integration. It’s the ability to create new secure and reliable ways for financial professionals to access information and complete work across the systems. The organizations that gain the most value from MCP will be those that use it to reimagine their workflows.