AWS re:Invent 2025 - Lessons from leaders: Turn AI agents into business value (API102)

Leveraging Agentic AI to Drive Business Value

Importance of Agentic AI Adoption

• Only 5% of companies are currently getting value from AI, according to an MIT study
• Agentic AI is a disruptive technology that will transform businesses faster than any previous technology
• Companies that start using Agentic AI now will gain significant competitive advantages

Key Lessons from Leaders

1. Start Small, Focus on High-Impact Use Cases

   • Begin with proven patterns in areas like customer service, back-office functions, and code development
   • Avoid complex "break the ceiling" use cases until foundational data and skills are in place
   • Prioritize use cases that can raise the floor and provide immediate value

2. Optimize for Outcomes, Not Processes

   • Agentic AI can shatter outdated processes and find new ways to deliver value
   • Focus on the desired business outcomes, not just improving existing workflows

3. Invest in Data Readiness and Quality

   • High-quality, accessible data is crucial for Agentic AI to deliver value
   • Data quality initiatives that couldn't be justified before may now have a strong ROI

4. Build a Robust Technical Foundation

   • Agentic AI requires the ability to access and integrate data from across the organization
   • Develop a platform engineering approach to enable Agentic AI to operate seamlessly

5. Adopt an Agentic Development Lifecycle (ADLC)

   • Traditional software development models are not well-suited for Agentic AI
   • The ADLC focuses on iterative optimization and learning, both in development and production

IBM's Agentic AI Transformation

• IBM has implemented Agentic AI across its business, including:

  • Automating back-office functions like HR, finance, and IT
  • Enhancing customer service and code development
  • Developing new offerings like IBM Consulting Advantage and a Agentic IDE

• Key results:

  • $4.5 billion in annual run-rate improvements through Agentic AI adoption
  • Increased productivity and innovation across the organization

Apollo Tires' Agentic AI Journey

• Challenges:

  • Siloed data across multiple systems and formats
  • Slow, manual reporting and analytics processes

• Solutions:

  • Consolidated data into a central data lake on AWS
  • Leveraged Amazon Bedrock and OpenSearch to enable natural language querying
  • Developed multi-agent frameworks to ensure data quality and accurate results

• Benefits:

  • Reduced report generation time from 15 days to 4-5 days
  • Enabled self-service data access and insights for business users
  • Improved data quality, consistency, and timeliness across the organization

Key Takeaways

1. Start small, focus on high-impact use cases to build momentum and demonstrate value.
2. Optimize for business outcomes, not just incremental process improvements.
3. Invest in data quality and accessibility as a foundation for Agentic AI success.
4. Develop a robust technical architecture to enable seamless Agentic AI integration.
5. Adopt an Agentic Development Lifecycle to support iterative optimization and learning.
6. Leverage Agentic AI to raise the floor and break the ceiling of business performance.