Back office ops · Production

DXC Technology builds LLM-powered AI assistant for oil and gas data exploration on Amazon Bedrock

The problem

Oil and gas companies need to discover new drilling sites and reduce time to oil, but their data is scattered across remote sites and offices, non-standard, and spans a wide variety of formats including spreadsheets, satellite images, GIS data, and industry-specific LAS files — making data exploration slow and costly.

Workflow diagram · grounded in source
1
User submits natural language query
trigger
“when a user asks "What API produced the most oil in 2010," a natural follow-up question would be”
2
Context-aware query rewriting
ai_action
“takes the user query and the conversation history and rewrites the user query with context that might be missing from it. We can also use this query-rewriting layer to directly translate or summarize previous responses, without having to…”
3
LLM router routes query to tool
routing
“The router analyzes the user query and routes it to the appropriate tool”
4
LLM generates Python code for data analysis
ai_action
“It uses the LLM's ability to write Python code for data analysis”
5
Semantic RAG search for document queries
ai_action
“The implementation for semantic content-based search relies on Amazon Bedrock Knowledge Bases. Amazon Bedrock Knowledge Bases provides a seamless way to implement semantic search”
6
Final answer returned to user
output
“Call Anthropic's Claude v2.1 model with the prompt to get the final answer”
Reported outcome

Data exploration tasks that previously took hours can now be achieved in just a few minutes, dramatically reducing time to first oil for DXC's oil and gas customers.

Reported metrics
Data exploration task durationhours can now be achieved in just a few minutes
Time to first oildramatically reducing
Reported stack
Anthropic's ClaudeAmazon BedrockAmazon Bedrock Knowledge BasesAmazon S3LangChainlasio
Source
https://aws.amazon.com/blogs/machine-learning/dxc-transforms-data-exploration-for-their-oil-and-gas-customers-with-llm-powered-tools?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Data exploration tasks that previously took hours can now be achieved in just a few minutes, dramatically reducing time to first oil for DXC's oil and gas customers.

What tools did this team use?

Anthropic's Claude, Amazon Bedrock, Amazon Bedrock Knowledge Bases, Amazon S3, LangChain, lasio.

What results were reported?

Data exploration task duration: hours can now be achieved in just a few minutes; Time to first oil: dramatically reducing (source-reported, not independently verified).

How is this back office ops AI workflow structured?

User submits natural language query → Context-aware query rewriting → LLM router routes query to tool → LLM generates Python code for data analysis → Semantic RAG search for document queries → Final answer returned to user.

Related back office ops cases