에이전트 AI가 금융 서비스 산업에 미치는 영향

This IDC Perspective examines selected areas in which agentic AI can impact financial services organizations. Agentic AI builds on the capabilities of GenAI and LLMs but adds additional abilities (e.g., planning and executing actions autonomously to reach objectives). These agents can use tools to support their activities and produce relevant outcomes or outputs; they can also cooperate with other agents and orchestrate tasks through the division of labor, as well as engaging in self-reflection and planning.These capabilities enable financial organizations to employ agents to run and orchestrate important use cases (e.g., personalization of the overall client relationship, including product management and pricing), deal with overdue debt collections, and automate the entire origination and underwriting process. "GenAI can truly elevate automation, but agentic AI takes this to a new level by bringing to life the latent knowledge hidden in large language models. The added value of autonomous agents is that they can make decisions by evaluating and improving their own performance, analyzing their outputs, decisions, and reasoning processes. This is truly a step toward augmenting non-human intelligence and enabling a machine to make independent decisions." - Associate Director Michael Yeo, IDC Financial Insights

Executive Snapshot

Situation Overview

• Definition and Concept
  • Agentic or Autonomous AI
  • Multi-Agent Systems
    • The Role of the Orchestrator Agent
    • The Role of the Specialist Agent
• Agentic AI Is Part of AI's Natural Evolution
• Use Cases: Challenges and Solutions
• The Benefits of Agentic AI for Financial Services
  • Benefits
• The Risks of Agentic AI for Financial Services
  • Risks
  • Governance and Risk Management for Agentic AI
  • Global Regulatory Reaction to Agentic AI in Financial Services
  • IDC Essential Guidance on Agentic AI
    • Platform Considerations for Agentic Workflows

Advice for the Technology Buyer

Learn More

• Related Research
• Synopsis