Gen AI-Powered Case Data Search
Industry:
Oil & Gas
Region:
AI-Powered Case Data Search Assistant
Technology:
Anthropic Claude LLM, AWS Bedrock, Python, Prompt Engineering
A leading U.S.-based provider of natural gas and natural gas liquids (NGL) transportation and storage services, serving a nationwide customer base. Field technicians routinely performed high-risk “hot work” activities such as welding and cutting on storage tanks and pipelines. A Business Process Management (BPM) platform was used to capture case data from field technicians and transmit it to a cloud data warehouse for further investigation and analytics.
Back-office teams faced significant challenges in managing field technician manpower:
- Difficulty in identifying specialized technicians for complex hot work tasks, resulting in project delays.
- Inefficient allocation of temporary worker agencies, leading to increased operational costs.
- Reliance on traditional keyword searches through investigation notes, observations, and comments proved ineffective for timely and accurate decision-making.
- User Interaction
- Query Classification
- Data Retrieval:
- Structured Queries: Processed through SQL queries against the Redshift DB.
- Vector Queries: Utilize vector embeddings stored in OpenSearch for semantic search.
- Response Generation
- Integration with Pega
The AI Assistant significantly improved decision-making speed and resource allocation. It enabled back-office teams to analyze case data in minutes instead of hours, leading to faster, more informed technician and agency selections. Additionally, semantic search enhanced technician-job matching, improving work quality, minimizing rework, and fostering more strategic partnerships.
