Vivix Vidros Planos Transforms Glass Manufacturing with Mendix Low-code and Amazon Bedrock

As the sole glass manufacturing plant to be 100-percent funded by Brazil, Vivix Vidros Planos is one of the most modern and sustainable flat glass factories in the world. With customers primarily in the construction and furniture industries, Vivix produces a variety of glass types using its own raw material. To keep its manufacturing operation on the leading edge, Vivix sought support and expertise from AWS Partner Mendix. Together, the two worked to digitize processes and incorporate generative AI through Amazon Web Services (AWS). In collaboration with Mendix, Vivix developed a series of applications, including the chatbot Virtual Engineer to accelerate knowledge sharing and productivity on the shop floor.

Established in 2014, Vivix has a team of 300 employees producing 900 tons of flat glass per day. However, Vivix faced significant challenges in its glass manufacturing operations, especially in rapidly training technicians while preserving operational knowledge. A significant obstacle was Vivix’s previous reliance on Excel and paper, with manual data consolidation feeding into SAP at the control layer. “In the past, Vivix didn’t have a solid foundation for our data architecture and information for industrial processes,” says Aristóteles Terceiro Neto, industrial transformation manager at Vivix. For instance, analyzing batches of glass involved manually tracking each step from layering, to cutting, to finalizing. Without digital solutions to house that information, it was time-consuming for engineers to solve issues on the line and address customer complaints.

Ultimately, Vivix wanted to increase the industrial teams’ productivity and evolve into a data-driven organization. The company sought a generative AI solution that could help address customer complaints, support employees on the factory floor, and give experts more time and energy to focus on structural improvements.

After considering more than 18 solutions, Vivix chose Mendix, an enterprise low-code platform from Siemens, to support its digital transformation. “What first stood out to me about Mendix was the possibilities. It’s more complete than other low-code alternatives and broadens the range of projects we can take on,” says Raissa Alencar, process and continuous improvement engineer at Vivix. After a three-year engagement, the glass manufacturer now has more than 25 Mendix applications supporting the entirety of its production and logistics processes. These applications are integrated with core systems like SAP S/4HANA and directly connected to production lines. Most of the applications are deployed on the Mendix Cloud, which runs on AWS, but some are also deployed on Siemens Industrial Edge devices. Mendix and its network also manage the ETL of operational technology (OT) data into AWS services and Mendix, integrating Vivix’s OT and IT data.

Mendix proposed that Vivix use Amazon Bedrock to build its smart chatbot, the Virtual Engineer. A fully managed service, Amazon Bedrock offers a choice of high-performing foundation models (FMs) from leading AI companies like Anthropic, Stability AI, and Cohere. Mendix included Claude-2, a large language model (LLM) designed to assist with various natural language processing tasks. Given the complexity of this project, Mendix relied on a network of partners to support the build and deployment of the Virtual Engineer. APIs, logic, and frontend components are managed by one partner while another handles the Retrieval Augmented Generation (RAG) application, ensuring employees receive accurate answers from the chatbot. Additionally, the deployment of an AWS Data Lake was instrumental in establishing the Industrial Data Fabric and the Unified Namespace (UNS), which provide the structured, semantically rich data foundation that the Virtual Engineer relies on.

The frontend of the Virtual Engineer is a chatbot, which is embedded into an existing Mendix application called Glass DNA—which serves as a central hub for production data, quality metrics, and operational insights. Glass DNA integrates seamlessly with the Virtual Engineer, providing a robust data foundation that enhances the chatbot's ability to generate accurate and actionable insights. It’s a critical application that helps Vivix’s quality and production teams track and trace products at the batch level, based on IT and OT data from several sources. Integrating the Virtual Engineer into an existing Mendix application that Vivix already uses ensures a seamless user experience and improves adoption. Vivix intends to apply this approach to other applications as well, gradually incorporating generative AI throughout its production processes.

Along with Amazon Bedrock, the Virtual Engineer leverages a suite of AWS services including the Amazon Relational Database Services (Amazon RDS), Amazon Elastic Compute Cloud (Amazon EC2), Amazon DynamoDB, Amazon Simple Storage Service (Amazon S3), Amazon Timestream, AWS Lake Formation, and Amazon Athena. These AWS services enable the Virtual Engineer to synthesize data from multiple sources, including Glass DNA, to generate accurate, actionable insights that support defect assessment, historical data analysis, and document information retrieval.

Mendix and AWS have empowered Vivix to enhance operational efficiency while expanding its strategic possibilities. "With Mendix and AWS, high-quality solutions and rapid development are a guarantee. They give me the space to think bigger and make a larger impact,” says Alencar. One of the standout innovations is the Virtual Engineer, which has transformed how Vivix handles customer complaints and production issues. When a customer complaint is received, the Virtual Engineer generates a detailed report to help the quality specialist validate the issue and develop an action plan. This streamlined process resulted in an 80 percent reduction in complaint resolution times and it turns insights into valuable feedback for improving production resilience. On the manufacturing floor, engineers—both junior and senior—use the Virtual Engineer to assess defects in real time and validate complaints. If a complaint is linked to a production issue, the tool enables teams to quickly diagnose the defect, drastically reducing the time needed to identify corrective actions. By integrating quality and process data, the smart chatbot has shortened resolution times from weeks to mere minutes.

Vivix’s strategic use of AI has improved its quality assurance process and accelerated responses to production issues by 85 percent. Leveraging generative AI models in conjunction with knowledge bases enriched with Vivix’s proprietary data, the Virtual Engineer accelerates technician training, allowing new hires to acquire critical operations and maintenance skills in months rather than years. “With generative AI, we’ve compressed our technician training process, enabling us to scale our production and quality efforts more efficiently, even as we onboard new talent,” notes Neto. Now, engineers of all experience levels have access to the same level of expertise when they need it most.

By innovating with AWS’s cutting-edge generative AI services, Vivix has preserved institutional knowledge while rapidly upskilling its workforce, optimizing operations, and positioning itself for sustainable growth. “We plan to integrate even more AWS services and develop additional AI-powered smart apps to create barrier-free data workflows. Our goal is to empower our team to use generative AI for better reporting and to gain fresh perspectives on our data,” adds Neto.

Vivix Vidros Planos, a Brazilian company, is renowned for its modern and sustainable flat glass production. With a diverse product range and a daily production capacity of 900 tons, Vivix serves the construction and furniture industries.