GAIA: an Agentic Artificial Intelligence System for Geothermal Field Development
Randy Harsuko, Zhengfa Bi, Nori Nakata · 2025
Read GAIA: an Agentic Artificial Intelligence System for Geothermal Field Development by Randy Harsuko, Zhengfa Bi, Nori Nakata (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-03852.txt.
Introduction
Randy Harsuko, Zhengfa Bi, Guodong Chen, and Nori Nakata introduce GAIA, an agentic artificial intelligence framework for developing and monitoring geothermal fields. Catalog evidence carried an earlier working title; the arXiv record gives the title used here. Decisions in this field must combine geology, geophysics, reservoir engineering, and operations data under time pressure.
GAIA joins three parts: an agent that plans tasks and queries knowledge bases, a digital twin holding physics and surrogate models, and a chat interface with visual controls and document retrieval. A curated benchmark of geothermal scenarios tests the system. By coupling physics models with agent reasoning, GAIA aims to automate full data workflows, not just single tasks.
Catalog record
| CREATOR | Randy Harsuko, Zhengfa Bi, Nori Nakata |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-03852
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
GAIA: an Agentic Artificial Intelligence System for Geothermal Field Development Thanks: Work done during the first author’s internship at Lawrence Berkeley National Lab Randy Harsuko, Zhengfa Bi, Nori Nakata Abstract Geothermal field development typically involves complex processes that require multi-disciplinary expertise in each process. Thus, decision-making often demands the integration of geological, geophysical, reservoir engineering, and operational data under tight time constraints. We present Geothermal Analytics and Intelligent Agent, or GAIA, an AI-based system for automation and assistance in geothermal field development. GAIA consists of three core components: GAIA Agent, GAIA Chat, and GAIA Digital Twin, or DT, which together constitute an agentic retrieval-augmented generation (RAG) workflow. Specifically, GAIA Agent, powered by a pre-trained large language model (LLM), designs and manages task pipelines by autonomously querying knowledge bases and orchestrating multi-step analyses. GAIA DT encapsulates classical and surrogate physics models, which, combined with built-in domain-specific subroutines and visualization tools, enable predictive modeling of geothermal systems. Lastly, GAIA Chat serves as a web-based interface for users, featuring a ChatGPT-like layout with additional functionalities such as interactive visualizations, parameter controls, and in-context document retrieval. To ensure GAIA’s specialized capability for handling complex geothermal-related tasks, we curate a benchmark test set comprising various geothermal-related use cases, and we rigorously and continuously evaluate the system’s performance. We envision GAIA as a pioneering step toward intelligent geothermal field development, capable of assisting human experts in decision-making, accelerating project workflows, and ultimately enabling automation of the development process. K eywords AI agents \cdot geothermal \cdot retrieval augmented generation 1 Introduction Geothermal energy is a renewable energy source that harnesses heat from the Earth’s interior for electricity generation and direct use applications. It has the potential to provide a significant portion of the world’s energy needs while reducing greenhouse gas emissions. However, geothermal field development is a complex process that involves multiple stages, including resource assessment, exploration, drilling, production, and reservoir management and optimization. Each stage requires specialized knowledge and expertise in various fields such as geology, geophysics, reservoir engineering, and environmental science ( Gensheng et al. 2024 ) . The complexity of geothermal field development presents challenges for decision-making, as it requires experts to integrate diverse data sources, interpret complex geological and geophysical information, and make informed choices under time constraints. To address these challenges, there is a growing interest in leveraging artificial intelligence (AI) and machine learning (ML) techniques to automate and assist in geothermal field development (e.g., He et al. 2020 ; Nakata et al. 2025 ; Gutiérrez-Oribio et al. 2025 ; Montes et al. ). Recently, AI agents have shown promise in various domains, including natural language processing (NLP), computer vision (CV), and robotics ( Chan et al. 2024 ) . AI agents, which are sometimes also referred to as "Deep Research Agents", are LLM-powered systems that integrate dynamic reasoning, adaptive planning, multi-iteration external data retrieval and tool use, and comprehensive analytical report generation for research purposes ( Huang et al. 2025b ) . They are meant to assist users in automating complex tasks, such as data analysis, literature review, and experimental design. Some examples of recent AI agents include OpenAI Deep Research ( OpenAI ) , Alita ( Qiu et al. 2025 ) , Pangu DeepDiver ( Shi et al. 2025 ) , etc. For a more comprehensive review of AI agents, please refer to Huang et al. 2025b . In the geoscience sphere, several LLM-based systems have been proposed. ChatClimate ( Vaghefi et al. 2023 ) is a web-based chatbot app built for querying climate-related information based on the latest IPCC report. It leverages the standard retrieval-augmented generation (RAG) pipeline to reduce hallucinations and provides more accurate and factual answers on climate. Pioneering LLM-based geological map analysis, Huang et al. 2025c promote PEACE, an agentic system with GPT-4o as a base model that excels in various geological map-understanding tasks. They also curate a benchmarking dataset for evaluating AI agents in geologic map understanding. AskGDR ( Weers et al. 2024 ) is the first LLM-based app introduced to the geothermal community. The chatbot-styled web app was trained on metadata from collections of geothermal datasets in the Geothermal Data Repository (GDR) and was designed to assist researchers in obtaining geothermal datasets that suit their needs. For seismic exploration applications, Kanfar et al. 2025 is the first to introduce an AI agent that specializes in building seismic processing workflows. Rather than manually writing Madagascar ( Fomel et al. 2013 ) scripts, the agent can automatically generate and execute the scripts based on user prompts, significantly reducing the time and effort required for seismic data processing ( Kanfar et al. 2025 ) . However, currently, there is still limited, if not any, research on agentic AI systems that handle a complete, end-to-end geothermal field development workflow. The objective of this work is to develop an AI-based system, GAIA – Geothermal Analytics and Intelligent Agent, for automation and assistance in geothermal field development. GAIA addresses the complexity and challenges of decision-making, which often requires integrating geological, geophysical, reservoir engineering, and operational data under time constraints. The system is designed to assist experts throughout the workflow, from data analysis and simulation to decision support and project automation. To the best of our knowledge, GAIA is the pioneering effort in building an agentic AI system for geothermal project assistance. 2 Methods Figure 1: GAIA Agentic RAG Workflow. Users interact with GAIA Chat interface by supplying prompts and supplementary data. The GAIA Main Agent will first think through the prompt and make a step-by-step plan to solve the given problem. If necessary, the system will fetch data from its knowledge base consisting of PDFs of papers/proceedings, tabular data, and/or external data sources (via web search and/or GDR). The collected data and the formulated workflow will be forwarded to GAIA DT for the actual processing through its set of tools. Finally, the Main Agent will formulate the final response based on the gathered information and multi-step analyses. GAIA comprises three components: GAIA Agent, GAIA Chat, and GAIA Digital Twin or DT, forming an agentic RAG workflow. GAIA Agent, powered by a pre-trained LLM, autonomously queries knowledge bases, executes domain-specific subroutines, and orchestrates multi-step analyses. GAIA DT integrates classical and surrogate physics models with interactive visualization tools, while GAIA Chat provides a web-based interface with features such as interactive visualizations, parameter controls, and context-aware document retrieval. Figure 1 illustrates the GAIA Agentic RAG workflow. Each of these components will be discussed in detail next. 2.1 GAIA Agent The core component of agentic AI systems is an AI agent, which is based on a pre-trained LLM. In the GAIA framework, we define a main agent that is responsible for planning, orchestrating, and executing tasks. Therefore, the minimum requirement of the pre-trained LLM is that it needs to understand instructions (e.g., via instruction tuning ( Peng et al. 2023 ) ) and generate human-like text. In this work, we use the open-weight model of Gemma 3 from Google Deepmind ( Team et al. 2025 ) , which is a 27B parameters LLM model that was trained on 14T tokens of text and image data. We specifically use the Gemma3-27B-IT (instruction-tuned) variant of the model, which is designed to follow user instructions and generate