ARTIFICIAL INTELLIGENCE
AI transforms a manufacturing startup
AI-driven solutions have powered the transformation of a Taiwanese startup, helping to manage their vast database of manufacturers and overcome critical inefficiencies in design, quoting, and data management. Thanks to a comprehensive AI-powered platform, the company has significantly improved their operational efficiency, as well as their customer satisfaction.
Operational challenges can limit the growth of startups and impact their market performance. To overcome these challenges, a Taiwanese firm partnered with VMO, a subsidiary of the ALTEN Group specialized in bringing innovative solutions to clients worldwide. VMO worked with the client to streamline searching, designing, and sourcing electronic components for their customers in the aerospace, automotive and automation industries.
Challenge: Dealing with market disparities while improving efficiency in designing and sourcing components
Solutions: A comprehensive web platform based on advanced AI tools such as LLM, OCR and automation
Benefits:
- Increased data consistency
- Faster design and quoting
- Improved customer decision-making and efficiency
- Heightened customer satisfaction
Key performance indicators
- 80% reduction in processing time compared to manual work
- 93% accuracy in reading information from complex drawings, 100% for simple drawings
- 88% reduction in time spent on searching
- 32% reduction in procurement costs
Three challenges, one solution
Overcoming the startup’s inefficiencies meant dealing with three critical challenges: component disparity, design inefficiency and quoting bottlenecks. The first challenge arose from the fact that the availability of electronic components across manufacturers was often inconsistent, making it difficult to access accurate and consistent data for component selection. This not only reduced company income but also led to dissatisfied customers. In design terms, the startup’s reliance on external tools for creating and adjusting technical drawings slowed the design workflow and introduced a high potential for errors, often delaying projects and complicating design adjustments.Finally, the manual generation of bills of materials and preparation of quotes was hugely time-consuming. Fluctuations in prices and shifting availability compounded this problem, making it difficult for the quoting system to keep up with market dynamics and causing delays and inaccuracies that impacted customer decision-making. VMO came up with a single solution that could address all of these challenges.
Transforming core processes with AI
The idea was to use AI-driven solutions to automate key processes, enhance data consistency and improve operational efficiency.
Data standardisation using large language models (LLM) helped address the disparity in data. An advanced AI system made it possible to analyse and standardise unstructured data from multiple component manufacturers, enabling the startup to offer users a single, consistent source for component specifications. In addition, a similarity search made it possible to identify equivalent components among various manufacturers, giving users more options.
For the design issues, automated document processing with optical character recognition (OCR) and object detection made it possible to streamline the process, converting technical drawings (such as PDFs and scanned documents) into structured, machine-readable data. Object detection algorithms were implemented to identify and extract relevant design elements from the drawings, drastically reducing the time and effort required to adjust designs. This eliminates the need for external design tools while minimizing human error for faster, more accurate design workflows.
The quoting process was animated and designed to include AI recommendations based on supplier availability and market conditions. This helps to reduce delays and errors in the quotes and bills of materials (BOMs) and is far more efficient than the manual process, enabling customers to make quicker and better-informed decisions.
A web application that does it all
All these AI-powered solutions were integrated into a new web application with a robust database housing 345,000 components sourced from 11 top manufacturers. The platform’s technical stack was built using NextJS and NestJS for the web application, MySQL and Redis for data storage, and Elasticsearch for fast querying, ensuring a high-performance environment. AI-driven technologies – including Python-based AI models, Amazon Textract, Google Cloud Vision AI, and in-house AI development – contribute to the platform’s functionality. Cloud platforms like AWS and GCP provide the necessary infrastructure for seamless scalability. Agile Scrum methodology allows for continuous development, ensuring iterative improvements and quick responses to user feedback.
Improved efficiency and customer satisfaction
The integration of AI solutions into the startup’s platform has yielded significant improvements across the board. By embracing AI technology, the Taiwanese startup has successfully transformed its platform, addressing long-standing inefficiencies. In turn, this comprehensive transformation has significantly improved the user experience and, as a result, customer satisfaction.
