AI in Financial services

AI-assisted loan underwriting workflow Emerging

Problem:

Commercial lending decisions require analysts to gather financial data, structure information, and draft memos across a multi-stage process that is slow and labor-intensive.

Capability:

AI agents for document analysis and workflow automation

Value:

Speeds up loan analysis and documentation while keeping human lenders in control of final decisions, reducing turnaround time on complex lending cases.

Barriers: Data security, integration with core banking systems, and maintaining human oversight for regulated credit decisions

Automated proposal and pitch document generation Emerging

Problem:

Wholesale banking relationship managers spend one to two weeks manually researching clients and drafting proposals, delaying client outreach.

Capability:

Multi-agent generative AI for research, analysis, and document drafting

Value:

Cuts proposal creation time from weeks to hours, freeing bankers to focus on client strategy and relationship-building rather than manual drafting.

Barriers: Data integration across internal systems and need for fact-checking to maintain accuracy and trust

LLM-based accounting fraud detection in financial filings Experimental

Problem:

Regulators and auditors need scalable ways to detect accounting fraud across thousands of securities filings, a task historically dependent on manual review.

Capability:

Document understanding and anomaly detection with large language models

Value:

Could enable faster, broader screening of financial disclosures to flag potential fraud cases for human investigation, improving audit and compliance efficiency.

Barriers: Current model accuracy is limited on this task per benchmark results, and regulatory acceptance of AI-flagged fraud signals remains unproven

Voice-native customer service agents for banking and insurance Emerging

Problem:

Call centers in banking and insurance face high latency and containment issues when handling customer inquiries and claims, driving up cost and customer frustration.

Capability:

Audio-native conversational AI combining speech recognition, turn-taking, and function calling

Value:

Reduces call latency and improves containment rates, lowering call center costs while maintaining natural, low-latency customer interactions.

Barriers: Integration with existing telephony and CRM systems, and trust/compliance requirements for handling sensitive financial data over voice channels