The client is a mid-sized construction engineering company delivering large-scale infrastructure projects.
Their engineering teams manage hundreds of DWG files across multiple drawing sets, where accurate instrument tagging is critical for compliance, coordination, and project delivery.
Manual processes limited engineering productivity.
Engineers manually searched DWGs for tags, managed scattered metadata, and maintained error-prone Excel indexes, leading to inconsistencies and version conflicts.
This created several operational risks:
The organization needed not just a search, but automated engineering intelligence — with validation, governance, and auditability built in.
Crunch-IS designed and delivered a cloud-native Instrument Engineering Agent capable of understanding DWG drawings, executing structured tasks, validating outputs, and escalating uncertainty to engineers.
The goal was clear:
To achieve this, we built a structured, event-driven architecture that transforms raw DWG files into validated, queryable engineering intelligence — starting with automated drawing synchronization.
Drawings are synchronized through AutoCAD Desktop Connector into Autodesk Platform Services (APS).
Every update automatically triggers processing pipelines, without manual uploads or rework.
We built a serverless AWS system that retrieves DWGs via APS APIs and extracts instrument tags using LibreDWG.
The system:
This replaces Excel tracking with a structured, governed instrument database.
Engineering data becomes queryable, auditable, and consistent.
Instead of implementing a simple RAG-based search tool, we deployed an agentic AI architecture capable of planning and executing multi-step workflows.
The Instrument Agent operates in four modes:
1. ASK — Natural Language Engineering Queries
Engineers can ask:
And the agent plans the query, uses structured database tools, cross-checks DWG context, and returns validated results with confidence scoring.
2. ACT — Execute Engineering Tasks
The agent performs operational workflows such as:
Example:
“Generate a compliance report for Line 302.”
The agent queries validated data, applies rule-based checks, generates the file, and provides a secure download link.
3. VERIFY — Built-In Validation & Confidence
Every result includes:
This prevents silent AI errors and improves engineering reliability.
4. ESCALATE — Human-in-the-Loop Governance
When ambiguity is detected, the agent:
This ensures compliance with engineering standards while maintaining operational control.
The solution uses Autodesk Platform Services, AWS serverless tools, and Amazon Bedrock.
The architecture includes:
The result is a scalable, governed AI execution layer built directly on engineering artifacts.
The AI-powered DWG Intelligence platform enabled the client to dramatically improve instrument data management. Engineering teams replaced time-consuming manual review with automated validation and governance processes, resulting in faster, more reliable access to instrument records.
The client achieved:
Engineering teams now retrieve validated instrument data in seconds, rather than manually searching across hundreds of drawings.
The system provides confidence scoring, governance, and full traceability — reducing operational risk while improving productivity.
Tell us what you’re building or where you’re stuck. We work with engineering and product teams on custom software, AI & ML, cloud infrastructure, DevOps, and UI/UX — from early scoping to long-term delivery. One conversation is usually enough to know whether we’re the right fit.
Email: [email protected]
We and our partners use technology such as cookies on our site to personalize content and ads, provide social media features, and analyze our traffic. Click “Accept” to consent to the use of this technology across the web.
