Xecta Cuts Well Scenario Build Time by 60% with Claude on Amazon Bedrock
At a glance
Xecta’s Production Advisor module is a powerful tool to investigate alternate configurations and operational settings for producing oil wells, but its form-based interface made scenario creation more time-consuming than it needed to be, especially for complex workflows, and created an opportunity to make the product easier for new clients to adopt.
New Math Data built a conversational AI agent powered by Claude on Amazon Bedrock and Amazon AgentCore that lets engineers describe, iterate, and validate complex well scenarios in plain English, reducing scenario build time by 60% and expanding platform adoption by 25%.
The production system shipped in under six weeks and is now in daily use by Xecta field engineers, with a follow-on build underway to extend the agent across additional Production Advisor capabilities.
Industry
Use Case
- Workflow & Process Automation
- Natural Language Processing
- Asset & Infrastructure Management
Solution implemented
- Claude on Amazon Bedrock as the reasoning engine — interprets engineer natural language, understands O&G domain terminology, and generates structured parameters for the Production Advisor with high accuracy
- Amazon AgentCore for agent orchestration — manages tool selection, multi-turn conversation flow, and execution across the full scenario-building lifecycle
- Custom GraphQL MCP tooling — reads the Production Advisor's live GraphQL schema and converts conversational intent into precise, validated mutations, ensuring the engine receives clean inputs at every step
- LangGraph and LangChain memory — maintains structured JSON state across conversation turns, tracking scenario parameters, unit conversions, and validation flags session-wide
- RabbitMQ and OTEL observability — handles output routing and end-to-end token cost tracking per query, confirmed in live operation
- Containerized Python library on Kubernetes — delivered as a self-contained, production-ready artifact deployable in Xecta's cloud-native environment with no additional infrastructure dependency
The value equation
- 60% reduction in time to build common well scenarios, freeing engineers for deeper analysis
- 25% increase in overall Production Advisor usage, expanding the tool's reach across the organization
- 50% faster client onboarding, helping new users reach first productive use in half the time
- Production-ready, containerized agent architecture extensible to new tools, skills, and parameters as the platform grows
- Three-deal AWS-funded engagement validates sustained business value at every stage from proof of concept to production
Company Snapshot
Xecta is an integrated technology company combining AI and physics-based modelling to help energy operators optimize production and asset management through user-friendly digital engineering tools.
Location
Texas, USA | Dubai, UAE
Customer Situation
Xecta builds hybrid physics and AI digital engineering tools for energy operators, with its flagship products, ProdX and NetX, giving teams the analytical depth to monitor and optimize well and network performance. The platforms process complex surface, subsurface, and production data field-wide, giving engineers the computational power to model scenarios that would take days to run manually. ProdX’s Production Advisor enables engineers to rapidly model alternate equipment and operational parameters on wells to time workovers, select optimal equipment, and optimize current operations.
Because of that analytical depth, the interface required users to navigate multi-field forms to define and iterate scenarios. Expert users found the process repetitive, particularly when building and testing multiple scenarios. Non-expert users — new hires, clients onboarding to the platform — faced a longer path to first productive use because they had to learn both the engineering workflow and the scenario configuration process. Every hour spent configuring forms was an hour not spent interpreting results or testing more scenarios.
This tension between tool sophistication and usability is a persistent challenge across the energy technology sector. As operators look to AI to compress decision cycles, the bottleneck is increasingly the interface, not the model behind it.
NMD Solution
NMD reviewed Xecta’s Production Advisor architecture and identified an opportunity to make the core analytical engine faster and easier to access through natural language. The opportunity was not to rebuild the platform, it was to add a conversational interface.
NMD selected Claude on Amazon Bedrock as the reasoning engine for its ability to interpret domain-specific O&G terminology and map it to structured system inputs with high accuracy. Amazon AgentCore provided the orchestration layer for managing tool calls, session context, and multi-turn dialogue. A custom MCP server wrapped Xecta’s GraphQL API so the agent could translate conversational intent into precise schema-valid mutations in real time. LangGraph managed conversation state and parameter tracking across turns. The full system was packaged as a containerized Python library and delivered Kubernetes-ready, with OTEL observability and per-query token cost tracking built in from day one.
What We Delivered
Within six weeks of deploying NMD’s conversational AI agent, Xecta recorded a 60% reduction in time to build common well scenarios and a 25% increase in overall Production Advisor usage. Client onboarding time fell by 50%, with new users reaching first productive use in half the time previously required. The agent shipped as a production-ready, Kubernetes-deployable artifact on day one – no additional infrastructure work required by Xecta’s team.
Production queries confirmed real Bedrock consumption, with per-query cost tracking visible to the engineering team through OTEL instrumentation. Customer sign-off was completed in April 2026. A follow-on production SOW is now being scoped to extend the agent across additional Production Advisor capabilities, building on the modular LangGraph architecture designed from the start to support new tools and skills without re-engineering the core system.
Ready to Transform Your Energy or Utility Business?
See how New Math Data can transform your energy or utility business with AWS-powered innovation.