Improving IDIADA’s Intelligent Chatbot Efficiency with Amazon Bedrock

Applus+ IDIADA, a recognized global partner in the automotive industry, has taken a significant step towards digital innovation with the launch of its Digital Solutions department in 2021. With over three decades of experience in design, engineering, and testing, the company has focused on optimizing its data strategies and offering customized solutions to its clients to address the technological transformation in the sector.

In this context, IDIADA has introduced its intelligent chatbot, AIDA (Applus Idiada Digital Assistant), which uses Amazon Bedrock, an artificial intelligence service that provides access to advanced models from various industry-leading companies. AIDA has been designed as a versatile assistant that assists in a wide range of tasks, from customer service to resolving complex technical issues, significantly expanding IDIADA’s team capabilities.

One of the standout features of AIDA is its classification system, which categorizes human interactions with the assistant. This mechanism allows requests to be divided into different types, directing each inquiry to a specific pipeline according to its nature, whether it is a simple query or more technical requests.

As interactions with AIDA have increased, the need to improve the categorization of these communications has become evident. Initially, user questions were quite simple, but the chatbot has adapted to more complex requirements, optimizing both the system’s efficiency and the allocation of computational resources, reducing response time and enhancing the user experience.

Collaboration with Amazon has been crucial in developing various classification approaches. Methods based on large language models, such as Anthropic’s Claude 3, have been implemented, in addition to exploring classic machine learning algorithms. This has led to a better understanding of interactions, achieving greater accuracy in request categorization.

The results of these initiatives have been promising. AIDA has demonstrated handling over 1,000 interactions daily, with a 95% accuracy rate in correctly assigning requests. This has translated into a 20% increase in IDIADA team productivity.

Looking ahead, the company plans to offer AIDA as a product that can be integrated into its clients’ environments, aiming to expand the impact of this innovative technology. The implementation of Amazon Bedrock underscores its commitment to developing responsible and secure artificial intelligence, ensuring that customers can confidently maximize the potential of generative intelligence.

via: MiMub in Spanish

Scroll to Top
×