Artificial Intelligence in Plastics Processing: How Do You Take the First Step?

Artificial intelligence is currently the most talked-about technology topic in the industry. But there is often a wide gap between the hype surrounding AI and its actual business value. Many medium-sized plastics processors know they need to address this topic. At the same time, the key question arises: Where to start—and which use case will deliver the greatest benefit for their own company?

For injection molding, extrusion, and compounding companies in particular, this is where it is decided whether AI will become a competitive advantage or a costly experiment. After all, AI for AI’s sake does not add value. It only becomes successful when it solves specific challenges: less manual labor, better planning, more consistent quality, earlier error detection, more efficient maintenance, and more informed decisions.

The first step isn’t choosing an AI tool, but rather identifying the business process and the specific benefits. Where are bottlenecks occurring today? Where is data missing? Where are decisions still being made based on experience rather than reliable facts?

In plastics processing, there are many useful starting points:

• Production planning: improved sequencing, reduced setup times, more stable capacity utilization
• Quality assurance: earlier detection of deviations, assistance with inspection reports and risk assessments
• Maintenance: predictive maintenance based on machine and sensor data
• Material usage: better control of formulations, recycled materials, and variants
• Sustainability: structured data for the circular economy, digital product passports, and ESG requirements

The Fraunhofer Cluster of Excellence Circular Plastics Economy (CCPE) emphasizes that AI in the plastics value chain can become a key driver of efficiency, quality, and circularity by 2030 —but that it can only realize its full potential on the basis of interoperable data structures, clear standards, and viable business models.

Many AI projects fail not because of the technology, but because of the data infrastructure. When recipes, bills of materials, batches, inspection characteristics, machine data, and quality information are stored in different systems, Excel files, or isolated databases, AI quickly gets stuck in the pilot phase.

For mid-sized plastics processors, this means: AI readiness begins with data readiness. An integrated Microsoft platform lays the foundation for this. COSMO ERP Plastics specifically extends Microsoft Dynamics 365 Business Central to meet industry-specific requirements for injection molding and extrusion—such as formula management, batch and lot management, variants, tooling planning, quality assurance, cost estimation, and traceability —and brings data together in a centralized location.

Through integration into the Microsoft ecosystem, additional components such as Microsoft Power BI, Microsoft Fabric, Microsoft Power Platform, Azure IoT, and Microsoft Copilot can be gradually connected. This avoids creating a new siloed solution and instead establishes a scalable data and process foundation for AI, analytics, and automation.

AI truly shines where real-time production data is available. In injection molding and extrusion, sensors, machines, and systems provide important information on runtime, downtime, temperatures, scrap, energy consumption, and maintenance status.

IoT-based production, predictive maintenance, and digital twins create transparency on the shop floor. In conjunction with COSMO ERP Plastics and the Microsoft ecosystem, this data can be used for better planning, predictive maintenance, quality analyses, and management decisions.

This makes AI not just an abstract concept, but a concrete reality: it helps identify patterns, detect deviations earlier, and continuously improve processes.

Microsoft’s approach offers businesses various entry points into AI. On the one hand, there’s in-app AI directly within Microsoft business applications (ERP, CRM) such as Microsoft Dynamics 365 Business Central, where Copilot—an AI-powered assistant— can simplify routine tasks, make data discoverable, and support users in their workflows.

On the other hand, the Microsoft Power Platform can be used to develop custom applications, automations, and AI-powered processes. For example, Copilot in Power Apps helps users create apps and data models using natural language.

This range of capabilities is particularly valuable for plastics processors: Microsoft’s out-of-the-box AI features can be deployed quickly, while specific use cases—such as quality inspection, formula data, machine status, or supplier processes—can be built specifically on top of the existing Microsoft architecture.

A concrete example illustrates how AI becomes tangible in an industrial context: COSMO CONSULT’s Autonomous Test Plan Agent is directly embedded in COSMO ERP Plastics and helps address a typical bottleneck in quality management.

When a supplier data sheet—often in PDF format—is received, the agent automatically creates a goods receipt inspection plan in the background. If necessary, it creates new inspection characteristics or inspection types and can optionally prepare additional data such as vendors, items, purchase orders, or goods receipts. QM specialists then review and approve the plan.

The added value does not lie in replacing human expertise. The agent handles repetitive data entry and reduces the workload on specialists, while the final decision deliberately remains with humans. It is precisely this “human-in-the-loop” approach that is crucial for mid-sized manufacturers: AI supports, accelerates, and structures—but quality, responsibility, and approval remain under human control.

Many plastics processors operate as suppliers to various industries—such as the automotive, construction, and consumer goods sectors, as well as highly regulated fields like medical technology. Depending on the customer segment, this can result in stricter requirements for the use of AI even for plastics processors.

In regulated GxP processes, specific requirements apply: AI models used must be traceable, validated, and controllable. Static models are permissible provided their functionality is not autonomously altered after development and validation. Self-learning models and large language models must not be used in critical processes without human oversight, as this complicates traceability and validation.

In short:

• Training data must be valid and traceable.
• AI models must not change uncontrollably in critical processes.
• Decisions must remain verifiable.
• Compliance must be considered from the very beginning.

This is precisely where integrated processes, audit trails, role- and permission-based concepts, and clean data structures pay off.

To ensure that AI doesn’t get stuck in the experimental phase, a structured approach is needed. Our AI consulting approach combines four phases—from the initial idea to scalable implementation. We’ll guide you through:

1. Inspiration: Identifying potential, evaluating ROI, and defining goals.
2. Foundation: Breaking down data silos, improving data quality, and establishing semantic foundations.
3. Automation: Integrating platforms, establishing governance, and empowering employees.
4. Evolution: Developing specialized applications and building competitive advantages.

This approach is particularly well-suited for mid-sized companies because it does not view AI as an isolated technology project. Instead, use cases, the data foundation, platform architecture, governance, and change management are developed collaboratively.

Conclusion: The right first step isn’t AI—it’s finding the right use case within the business process

Artificial intelligence can help plastics processors become more efficient, transparent, and resilient. However, success depends on whether AI is embedded in a comprehensive process and data strategy. Without good data, clear goals, cloud readiness, governance, and acceptance within the business units, AI remains a patchwork solution.

To address this, COSMO CONSULT combines industry expertise in plastics processing with the Microsoft platform and integrated solutions such as COSMO ERP Plastics. This creates a robust foundation for ERP, analytics, IoT, Copilot, agents, and custom AI applications—not as an end in itself, but with measurable benefits for production, quality, IT, and management.

Would you like to read more about artificial intelligence in plastics processing?

Click here to read the press article by our Industry Manager Torsten Harnack in PlastXNow:

Artificial intelligence is not a utopian dream, but the next step