Turning Scattered Information into Searchable Intelligence
Giving teams faster access to transformer specifications and
enterprise knowledge, with Claude supporting architecture, documentation and development.
Overview
An electrical power and distribution provider needed a faster way to find transformer specifications and company information scattered across documents and emails. 10Pearls designed and built a multi-agent AI platform that answers questions in natural language, bringing critical knowledge within easy reach.
Previously 30+ seconds
Previously 15–20 second
On architecture & docs
Challenge
Finding information meant working across two different environments. Transformer specifications were held in structured data, while company knowledge was spread across documents and emails. Each specification search took more than 30 seconds, and every availability synchronization took around 15 to 20 seconds.
The right platform was expected to accommodate both kinds of searches. Equipment requests required filtering against specification fields while company information queries relied on finding and interpreting relevant material within documents and correspondence. Users needed access to both without navigating separate systems, with permissions controlling what they could see based on their role.
Building that experience also required coordination across applications, data sources and access controls, supported by clear requirements, architecture and technical documentation.
Solution
10Pearls gave teams a single place to find transformer specifications and company knowledge by asking questions in everyday language. The platform automatically directs each question to the appropriate specialist AI agent, so users can find the information they need without knowing which system holds it.
Equipment search is handled by an agentic system called SpecAI. Users describe the transformer specifications they need in natural language. SpecAI uses text-to-SQL to translate those requests into database queries, with spaCy supporting parameter extraction.
Enterprise knowledge is managed by a different system named KnowledgeBase. It’s a separate retrieval-augmented generation (RAG) agent that answers questions using company documents and emails. Content is ingested from SharePoint and a dedicated mailbox folder, processed through Azure Document Intelligence, then indexed in Azure AI Search.
The build supported shared access across applications. SpecAI, KnowledgeBase and EvalAI, the third application, are all available through a common portal. Azure AD single sign-on simplifies access, while role-based permissions control access to information and functionality.
Claude supported the engineering work behind this platform. The team used Claude Opus and Claude Sonnet for requirements analysis, preparing client discovery questions, Azure solution architecture, architecture diagrams and technical documentation, including README files. Claude Code, used within VS Code, supported codebase setup and development.
The platform gives teams faster access to equipment specifications and company knowledge through a single, permission-controlled portal, reducing time spent searching across disconnected systems. By bringing specialized AI agents into a shared experience, we simplified information access while preserving enterprise access controls, helping teams make informed decisions and respond to business needs more quickly.
Results & Benefits
Transformer specification searches now return in 1–5 seconds, compared with more than 30 seconds previously. Availability synchronization takes no more than 5 seconds, compared with 15–20 seconds before. Between 100 and 150 active users access the platform’s three AI applications through a single sign-on portal.
These platform improvements sit alongside a separate delivery benefit. Across architecture and documentation activities, the team estimated that using Claude saved 30–35% of the time required. The engagement combined faster access to operational information with AI-assisted engineering across requirements, design, documentation and development.
Customer
Electrical power and distribution provider
SERVICES
- Agentic AI Systems
- AI Integration
- Data Engineering
- Product Engineering
Tech stack
- Azure AI Search
- Azure Document Intelligence
- Microsoft Entra ID (formerly Azure AD)
- spaCy
- SharePoint
Developed with:
- Claude Opus
- Claude Sonnet
- Claude Code (in VS Code)
Related case studies
Energy
Transforming RFQ Configuration & Quoting with AI
Explore how 10Pearls used AI to automate transformer RFQ configuration and quoting, validate requirements, reduce errors, and accelerate accurate quote generation.
Energy
AI-Powered Transformer Operations Transformation
Discover how 10Pearls helped an energy provider build an AI-powered digital foundation, digitize operations, reduce cycle times, and connect enterprise knowledge.
Energy
Automated Transformer Drawing Generation
Discover how 10Pearls automated transformer drawing generation using evaluation data to create accurate, customizable 2D diagrams and reduce manual drafting time.
Energy
LiDAR Transformer Assessment Solution
Learn how 10Pearls leveraged LiDAR technology to digitize transformer assessments, streamline field inspections, improve measurement accuracy, and reduce evaluation time.
Energy
AI-Assisted Transformer Inspection Solution
See how 10Pearls streamlined transformer inspections with AI-driven evaluation workflows, improving measurement accuracy, reducing manual effort, and accelerating field assessments.
Energy
Enterprise Knowledge & AI Search Solution
Explore how 10Pearls developed an enterprise AI search solution that unifies business knowledge across documents, emails, and SharePoint for faster, smarter information retrieval.
Energy
AI-Powered Transformer Inventory Recommendations
Discover how 10Pearls built an AI-powered transformer inventory recommendation engine that improves inventory matching, accelerates decisions, and streamlines operations.
Energy
AI-Driven Inventory Matching for Smarter Operations
A leading electrical power and distribution provider faced inefficiencies in manually matching transformer orders to available inventory. 10Pearls developed an AI-powered application that automated order matching.
Energy
EV Charging Management Cloud Platform
A cloud-native EV charging platform scaled faster, improved reliability, and delivered the real-time performance needed to support growing demand.
Energy
Revenue Assurance Platform on AWS
Explore how 10Pearls replaced costly SaaS with a custom AWS invoicing and analytics platform, cutting billing from days to 2.5 seconds and flagging $150K in under-billing.
Ready to get started?
Contact us today to learn how we can transform your products or ideas into scalable solutions that drive business value and growth.