AI Strategy for Process Engineering at Continental Barum

In collaboration with experts from Continental Barum, we developed a strategy for deploying artificial intelligence in tire production management for the Process Engineering division.

Why Process Engineering

Process Engineering is one of the key divisions of the entire company. It sets up and fine-tunes production processes to achieve maximum efficiency and quality. It covers everything from product specifications through machine data analysis to the production run itself. The decisions made here shape the success of the whole plant. That is exactly why we greatly value the trust with which the director of this division, Tomáš Vilímek, approached us. Our task was to design solutions for this essential part of production.

The brief: no "AI for AI's sake"

From the very beginning it was clear that the strategy was not supposed to be merely a collection of trendy ideas with no real application. The goal was to find specific solutions that:

• reduce the scrap rate,

• speed up the response to defects,

• eliminate manual work with data across the entire production process, from preparation to final inspection.

Our approach: first understand the production, then design AI

Before we proposed a single solution, we spent weeks in deep-dive interviews with technologists and managers across all sections and in the analysis of real production data - product specifications, machine data, defect history.

This step was essential. Every finding in the resulting strategy is backed by a specific session or specific data, not by a theoretical model. As a result, the strategy reflects the department's actual operational needs, not a generic idea of what "AI in manufacturing" might look like.

The result: a portfolio of 26 use cases

The output is 26 specific use cases. From reporting improvements through predictive models to autonomous AI agents capable of independently responding to deviations in production.

We rated each use case along two axes: deployment difficulty and size of the benefit. This produced a natural implementation order:

1. Quick wins - low-effort solutions that deliver value almost immediately.

2. Strategic bets - more complex projects with the greatest financial impact.

The benefit of each use case is genuinely calculated against a specific production baseline, not estimated.

"Thanks to the collaboration with Blogic, and despite the high sophistication and complexity of our production processes, we obtained a concrete and understandable AI strategy built on real use cases of our division, not on theoretical promises. The team helped us create a vision ranging from simpler use cases based on automation and machine learning all the way to an agentic approach and a path towards agents that orchestrate and manage other agents."

Tomáš Vilímek, Director of the Process Engineering division

Strategic goal: data as a tool of prevention

Production data today is generated in volumes no human can go through manually. AI can - in millions of records it uncovers anomalies before they become a problem, or it can predict them in advance.

We therefore assembled the entire use case portfolio so that it gradually moves the work with data through three phases: from detection (what happened, in real time), through prediction (what will happen, ahead of time), to prevention (what will not happen, because it was prevented).

This shift is not just about technology - it is a change in the approach to quality management. Instead of repeatedly firefighting individual problems, it is about understanding what systematically distinguishes good production from scrap production, and managing the processes accordingly.

Use cases are moving into practice

One example of using complex production data is evaluating the correct course of the curing (vulcanization) process. Either already during the production process or retrospectively for additional quality confirmation. Machine learning models and intelligent data processing deliver results to technologists and quality engineers in a flash. It is a concrete example of the principle the whole strategy is built on: detection → prediction → prevention.

In addition, other use cases in the portfolio are moving forward as well. A model for classifying defect types from photographs is now in validation; the goal is for AI to be able to evaluate all available data and take over part of the decision-making that technicians perform today. AI support for mold design is heading into pilot operation; it will accelerate the iterative development of tools and fixtures thanks to a combination of machine learning on data from drawings and measurements.