Turning Design Diagrams into an Answerable Knowledge Base
Industry: Construction & Engineering Design
Engagement: Data extraction, agentic architecture & delivery
The Challenge
Construction and fabrication projects begin as dense CAD design diagrams that encode every component of a build. Turning those drawings into a usable bill of materials (BOM) is typically slow, manual, and error-prone – and once produced, that data is often locked inside files and tables that non-technical users can’t easily query or aggregate.
The Solution
Acxelin Agentix built a two-part solution: an extraction engine that reads construction design diagrams, identifies components, and organizes them into a clean, positioned bill of materials and an agentic AI data layer, built on GraphQL and the Model Context Protocol (MCP), with a natural-language chatbot that makes that multi-tenant data instantly answerable to any question, including complex aggregations, with governance and validation built in.
Business Impact
- Automated design-to-BOM – structured, positioned bills of materials generated directly from design diagrams, faster and more consistent than manual extraction
- Data made answerable – any user can query multi-tenant project data in plain language, including complex aggregations
- Trustworthy answers – deep validation and grounding keep responses accurate, explainable, and paired with helpful suggestions
- Secure by design – tenant isolation, access control, and governance enforced across every query
- A reusable foundation – the validated data and agentic tooling are built to feed downstream platforms and future AI workflows
Technologies
CAD/design-diagram extraction · GraphQL API · Agentic AI (Model Context Protocol) · Natural-language chatbot · Multi-tenant security & governance