Processing and cleaning production data is often a complex task, but as improvements are scaled across the organization, this process becomes a top priority.

The Importance of Data Integration

Large data warehouses collect information from various sources distributed throughout the enterprise. The broader the range of data coming from production and logistics processes, the more effectively artificial intelligence can identify patterns and generate forecasts.

Isolated data collection from separate sources leads to fragmented data silos. To bridge the gaps between such systems, the Brazilian glass manufacturer Vivix implemented the Mendix low-code AI application development platform, which enabled the integration of different data repositories.

Results of Implementing a Unified Data Architecture

The introduction and cleansing of a unified data architecture allowed Vivix to achieve the following results:

• Reduced time to resolve production issues by 85%
• Saved 6,000 work hours per year
• Increased furnace energy efficiency by 5%
• Lowered maintenance costs by $1 million in 2025

Accelerated customer request processing became possible thanks to the unified data architecture: response times dropped from two to three weeks to just two to three days. This enabled faster decision-making, reduced operational costs, and ensured process stability.

Organizational Changes and Approach to Digital Transformation

The creation of an industrial transformation department was initiated by company leadership, providing broad opportunities for experimentation and building a portfolio of solutions, rather than limiting efforts to a single pilot project. The company also introduced the role of value manager to track project outcomes.

Vivix worked with modern automation systems, which facilitated the creation of the necessary data foundation. The main challenges arose from fragmented business levels, not technological limitations.

Agile Methodology and Reuse of Architecture

The unified database supports experiments with agent-based AI, digitalization, predictive maintenance, and robotics. The more data points available, the easier it is to identify directions for innovation. Even unsuccessful projects contribute to data accumulation for future analysis.

The architecture and established foundations are reused for projects in logistics, maintenance, production, and safety. An agile methodology is applied: if a project does not deliver the expected value, it is stopped and resources are redirected to other initiatives.

As a result, the company gained the ability to efficiently utilize any available data, fostering the development of new solutions and opening up opportunities for further innovation.