AWS and TwinKnowledge Use AI for Construction Drawing Analysis

Amazon Web Services (AWS) has partnered with TwinKnowledge, an AI company focused on the architecture, engineering, and construction (AEC) industry, to develop a new method for automated construction drawing analysis. The solution leverages computer vision and large language models (LLMs) to streamline the drawing review process, aiming to reduce errors and enhance project efficiency.

TwinKnowledge, which seeks to decrease errors in construction documents using AI, faced scalability challenges in deploying its computer vision platform across clients. The AWS Prototyping and Cloud Engineering (PACE) team assisted the company in overcoming these limitations by utilizing Amazon SageMaker's MLOps capabilities and building a robust pipeline for data ingestion and validation.

The new system integrates computer vision's ability to identify visual elements within drawings with the understanding capabilities of large language models, enabling efficient processing of complex blueprints. This allows for the review of thousands of drawings within days while maintaining high accuracy.

The collaboration has established a foundation for enhanced search functionalities and AI-driven insights within the construction sector. The solution is intended to significantly augment the human review process and expedite the identification and correction of discrepancies in building projects.