Vivix is a float-glass manufacturer in Brazil producing about 900 tons a day. Its problem will sound familiar to plants in Vietnam: rich operational data spread across systems that never talked to each other, and incident resolution that depended on a handful of experts and a lot of manual effort.
Before the change, a production issue took four to five hours to resolve and a customer complaint took five days. The knowledge existed in experts’ heads; the coordination to act on it did not.
What they did
Within a year the Vivix team built 17 Mendix applications connecting production data, quality systems and institutional expertise into one environment. On top of that they built the Virtual Engineer: an AI agent embedded in existing workflows that gives anyone on the team access to the context a few experts used to hold.
Results
Production issues resolved in minutes instead of hours. Customer complaints closed in under a day instead of five. 6,000 hours of manual work recovered in the first year, an 80% reduction in complaint resolution time and 85% in production issue resolution time.
Mobino’s take
The AI did not win because it was a better model. It won because it had shared context, was designed into the workflow from the start, and the handoff between agent and human was intentional. Those are the three architectural gaps we keep talking about: context, execution and governance. With Vietnamese enterprises we start from one painful process, connect the data first, then place the agent.
Source: Vivix customer story published by Mendix (Siemens), 2025–2026. Vivix is a Mendix platform customer, not a Mobino project.