AWS re:Invent 2025 - How Baker Hughes is Driving Energy Innovation with AWS AI (AIM347)

Summary of AWS re:Invent 2025 - How Baker Hughes is Driving Energy Innovation with AWS AI (AIM347)

Energy Industry Challenges and Opportunities

• The energy industry has been pursuing the vision of the "digital oil field" for years, driven by rapidly increasing global energy demands.
• The industry generates massive amounts of data (15 petabytes per drilling rig, 1,800 rigs globally), but extracting meaningful insights from this data remains a key challenge.
• While the world is not running out of oil, the focus has shifted to extracting harder-to-reach and more expensive oil, driving the need for new technologies.
• Emerging technologies like large language models (LLMs), agent-based systems, and machine learning are seen as transformative for the industry.

Baker Hughes' Approach to Energy Innovation with AWS AI

Lucipa: Baker Hughes' Digital Technology Program

• Baker Hughes has been on a journey of digital transformation since the early 2000s, starting with physics-based models and classical techniques, then adding machine learning and AI, and now integrating agent-based systems.
• The Lucipa program focuses on:
  1. Accessing and contextualizing data
  2. Automating workflows and business processes
  3. Delivering outcomes and insights to customers
• This approach has evolved from simple plotting functions to more advanced reservoir simulation, nodal analysis, and machine learning models.

Architecting Agent-Based Solutions for Energy

• Baker Hughes is taking a flexible, customer-centric approach to integrating agent-based systems into their Lucipa platform.
  • Agents are viewed as partners and teammates, not replacements for human experts.
  • The architecture allows for seamless integration with customers' existing data sources, tools, and business processes.
  • Orchestration agents coordinate the interactions between various specialized agents and human experts.
• A key use case involves reservoir monitoring agents detecting events (e.g., rising oil-water contact) and triggering recommendations from Baker Hughes' pump optimization agents.
  • These recommendations are validated by human experts before being implemented by the customer.

Lessons Learned and Best Practices

• Data quality is foundational - garbage in, garbage out.
• Explainability of agent recommendations is crucial, especially for highly technical users.
• Adaptability to customers' heterogeneous environments and workflows is essential.
• Governance and cost controls are important, especially in heavy industries with high-stakes consequences.

The Future of Energy Innovation with AWS AI

• Baker Hughes sees agent-based systems as a game-changer for scaling their digital solutions and onboarding customers.
• By weaving agent-based systems into their Lucipa platform, they aim to automate and customize workflows, tailoring solutions to individual customer needs.
• This approach is enabled by open-source projects like Energy Agents, sponsored by AWS, as well as other AWS technologies for data management and governance.
• The goal is to leverage agent-based systems to drive impactful outcomes for energy customers, going beyond manual implementation efforts.