LLM-agents for data retrieval and analysis
During the Fall 2024 to Spring 2025, we collaborated with Solution Seeker to develop an interactive chat application powered by a Large Language Model (LLM) agent, aiming to simplify and accelerate data exploration and analysis for domain experts.
The application was built using Plotly Dash and integrates an LLM-based agent that interprets natural language prompts to generate code and database queries. Depending on the prompt, the agent can retrieve relevant data and produce visualizations or statistical analyses, offering a hands-free interface for interacting with complex datasets.
As part of the development, we investigated different approaches to agent architecture, including OpenAI’s Agents API and the Model Context Protocol (MCP).
Our final delivery consisted of a proof-of-concept application using OpenAI’s Agents API as well as documentation including the capabilities of the different agent backends, trade-offs between control, flexibility, and ease of use, and suggested future work.
-
Solution Seeker is a pioneering technology company specializing in artificial intelligence and machine learning solutions for optimizing oil and gas production. Since its founding, Solution Seeker has provided real-time, data-driven decision support tools aimed at enhancing efficiency, reducing emissions, and maximizing value for energy operators worldwide, with its headquarters located in Oslo, Norway.