Get in touch
Data discrepancies often require time-consuming manual investigation. This case study details how Master of Code Global helped a prominent US-based Energy company with an innovative solution: a Data Discrepancy Reconciliation Tool – a web application featuring an Agentic AI to expedite detection and resolution of the discrepancies. Powered by Generative AI, it provides natural language explanations for discrepancies in the ticket data, streamlining a complex resolution process. Let’s dive deeper to discover more about this project and its benefits!
Our client is a US-based Energy company, dedicated to managing energy assets across key production regions. They continuously search for innovative approaches to refine key processes and maintain a competitive edge in the industry, while maximizing ROI.
Inconsistencies between water and oil readings are an inherent challenge in the sector. For our client, these differences were creating substantial operational hurdles. Operations personnel were dedicating considerable time to manually reconciling data from various sources. It was slowing down processes, escalating operational expenses, and increasing the likelihood of reporting inaccuracies.
This manual approach was not only time-consuming but also susceptible to human mistakes, further exacerbating the problem. The firm acknowledged the necessity for a technological solution to automate this crucial process and liberate its workforce to focus on more strategic initiatives. The core issue was the tedious manual reconciliation.
Master of Code Global developed a cutting-edge, standalone web application: a Data Discrepancy Reconciliation Tool powered by Generative AI.
The solution automates the detection and explanation of inconsistencies between water and oil readings, streamlining the resolution process. The app offers an intuitive, user-friendly interface that displays data differences in a clear, visual format. In essence, the system utilizes advanced mathematical formulas and pre-defined thresholds to automatically identify variations between data origins.
Such a solution is an important breakthrough in applying artificial intelligence in order to refine and improve workflows in the Energy industry. By automating critical processes and providing understandable, clear, and short summaries of what is going on, Master of Code Global empowered the US company to increase their overall performance, lower the spendings and benefit from well-thought decision-making.
At Master of Code, we go beyond just developing AI solutions—we partner with our clients to ensure every implementation is strategically designed for success.
Our team played a pivotal role not only in building the solution but also in leading an in-depth Discovery & Design phase. This phase is the foundation of any successful AI implementation, allowing us to align business goals with technical feasibility. Leveraging our proven AI Discovery Framework, we conducted a series of structured design and technical workshops to:
✅ Define clear requirements tailored to the client’s unique needs
✅ Establish a robust architecture that ensures scalability and performance
✅ Set success criteria and measurable AI performance metrics to track impact
By taking this holistic approach, we ensure that AI solutions are not just deployed—but deliver real, measurable value. This methodology has become a key differentiator in helping our clients maximize the potential of AI-driven transformation.
Beyond simply flagging inconsistencies, this application incorporates a sophisticated, autonomous AI assistant at its core. This key component, built upon a foundation of Generative AI, acts as a virtual expert, proactively communicating data discrepancies in the chat. Instead of merely presenting a list of mismatches, Agentic AI offers concise, easily understood explanations in plain language, pinpointing the likely root cause of each discrepancy.
This is achieved through a combination of advanced techniques. The system utilizes Azure Cognitive Services with the GPT-4o-mini LLM, fine-tuned for this specific task, and is built with help of LOFT open-source LLM orchestrator framework. This allows the AI agent to not only understand the mathematical and logical rules governing data validity but also to translate complex patterns into human-readable narratives.
The generative capabilities analyze information, considering the applied discrepancy rules (like volume variance, source verification, BS&W checks, etc.), and then construct a clear explanation, such as: “The reported offload volume differs from the hauler volume by more than the acceptable 2.7% threshold,” or “The reported source does not match the offload report; GPS verification is recommended.“
This level of automated, intelligent analysis significantly reduces the cognitive load on haulers and operational staff. They no longer need to spend hours poring over spreadsheets and comparing data points; instead, they receive immediate, actionable insights to quickly resolve issues and focus on other critical tasks. The AI assistant effectively acts as a knowledgeable partner, accelerating the reconciliation workflow and improving overall operational efficiency.
Secure User Login: Distinct access portals for Haulers and Administrators, guaranteeing data protection and role-based access control.
Automated Data Ingestion: Scheduled file ingestion from Snowflake DB ensures the system consistently operates with the latest information.
Proactive Email Notifications: Alerts haulers to inconsistencies requiring their immediate attention, promoting rapid resolution.
Automates the inconsistency detection process, eliminating hours of manual data examination.
Minimizes human mistakes in the reconciliation procedure, resulting in more dependable reporting and decision-making.
Frees up haulers and operational personnel to focus on essential duties, rather than laborious data reconciliation.
The AI Agent delivers clear, natural language explanations for inconsistencies, enabling users to swiftly understand and address issues.
Seamless integration with Snowflake DB and the email notification system improves the entire data management operation.
Constructed on a robust, cloud-based infrastructure (Azure) to guarantee scalability and safety, adhering to industry data privacy regulations.
Verifies source/destination using geofencing, which helps prevent errors.
Ensures the correct load type is selected based on cost centers, avoiding misallocated expenses.
NestJS (TypeScript)
LOFT (open-source LLM orchestrator)
Azure Cognitive Services
React (TypeScript)
Azure Cloud Services
Dmytro Hrytsenko
CEO
John Colón
VP of Global Sales
Ted Franz
VP Sales & Partnerships
