The Abu Dhabi National Oil Company (ADNOC) and AIQ recently announced the successful proof-of-concept trial of ENERGYai, the world’s first-of-its-kind agentic artificial intelligence (AI) program tailored for the energy sector. ENERGYai integrates a 70-billion-parameter large language model (LLM) with more than 50 years of ADNOC’s knowledge and petabytes of its proprietary data.
The 90-day proof-of-concept trial demonstrated that ENERGYai’s agentic AI—AI “agents” that are trained in specific tasks across the energy value chain—can deliver significant improvements in the pace and accuracy of upstream exploration through rapid, precise, and detailed seismic survey analysis, alongside relevant, actionable insights to support production optimization at ADNOC’s existing wells. ADNOC reports that ENERGYai was able to deliver its insights in intuitive natural language, enabling engineers to interact with it effectively and easily.
ADNOC said the trial resulted in a 70% improvement in accuracy in major seismic interpretation aspects and significant improvements in advanced reservoir monitoring and anomaly detection.
“The successful completion of this proof of concept for ENERGYai has shown extremely promising results and has confirmed the potential of the solution to be a powerhouse for value creation and sustainable energy production,” said Musabbeh Al Kaabi, CEO for ADNOC Upstream. “Building on this initial achievement, ENERGYai will leverage petabytes of data to better empower our people and future-proof ADNOC, as we work to become the world’s most AI-enabled energy company.”
ADNOC said the ENERGYai trial also showed promising results in enhancing data quality, vastly improving the reliability and usability of operational data inputs by detecting errors, standardizing formats, and enriching data sets.
“This milestone was made possible through close alignment between ADNOC subject-matter experts, the expertise of 100+ specialists, and the establishment of robust and secure AI infrastructure,” said Magzhan Kenesbai, AIQ’s acting managing director. “These factors lay the foundation for even greater accomplishments in the forthcoming phases.”
Following the proof of concept, the first operational, scalable version of ENERGYai is expected to be completed in the first half of 2025. This version is expected to include five fully operational AI agents covering tasks within subsurface operations and to be test-deployed across a number of upstream assets, with plans to scale its application to thousands of additional wells.