AI in Plastics & Packaging Manufacturing: From ERP Data to Earlier Decisions

In the plastics and packaging industry, a delayed decision can cost a business more than expected.

A sudden surge in orders can put pressure on production schedules. Fluctuating raw material prices can erode profit margins. Equipment operating abnormally can disrupt delivery schedules. And by the time reports are compiled, the problem may have already occurred several days earlier.

The problem is not always a lack of data. Many manufacturers already have ERP systems, accounting software, production data, and management reports. However, data may still be scattered across different systems, updated too slowly, or underutilized when it comes to identifying risks.

AI in plastics manufacturing can help businesses move from reviewing what has happened to identifying what might happen next. When combined with relevant ERP data, AI and predictive analytics can help managers detect anomalies, assess trends, and make more proactive decisions.

But which business challenges can AI actually address? And what do manufacturers need to prepare to turn this technology into business value?

1. Why Do Plastics & Packaging Manufacturers Need to Make Decisions Earlier?

Plastics and packaging manufacturing involves specific operational challenges that make effective management more complex.

Manufacturers must manage multiple raw materials, product specifications, production formulas, customer orders, and quality requirements. At the same time, they must balance machine capacity, labor, inventory, and delivery schedules.

These factors are closely interconnected. Even a small change in the production plan can affect multiple departments.

For example, when a customer changes an order quantity, the planning team needs to reassess material requirements. Procurement must determine whether suppliers can meet the revised demand. Production needs to review machine capacity and operating schedules. Meanwhile, finance must evaluate the impact on costs and profitability.

If each department works with a different data source, decision-making can become time-consuming.

For CEOs, the challenge goes beyond the efficiency of individual departments. The more important question is whether the business can understand the overall impact of a decision before it affects costs, cash flow, or customer commitments.

Three common challenges stand out.

Data tells businesses more about the past than the future

Revenue, inventory, and production reports show what has already happened. However, they may not help businesses identify potential material shortages, delivery delays, or declining profit margins early enough.

Risk signals appear across multiple systems

A change in material consumption, a revised customer order, and unstable equipment performance may all be related to the same underlying problem.

If data is not properly connected and analyzed, these relationships can easily go unnoticed.

Managers spend too much time consolidating information

Instead of focusing on evaluating options and making decisions, management teams may spend considerable time reconciling figures, investigating causes, and requesting updates from different departments.

This is where ERP, data analytics, and AI can work together to close the gap.

2. ERP and AI: Different Roles That Complement Each Other

A common misconception is that implementing AI means a business can automatically solve every operational problem.

In reality, AI cannot independently produce reliable data when business processes are poorly controlled. Nor does every business challenge require AI.

ERP and AI serve different but complementary purposes.

ERP provides data and business context

A suitable ERP system helps businesses connect activities such as sales, procurement, inventory management, production planning, finance, and accounting within a unified management platform.

Depending on the system and implementation scope, ERP can provide data on:

• Customer orders, demand, and delivery schedules.
• Raw materials, bills of materials, and inventory levels.
• Production plans, actual output, and resource utilization.
• Material costs, manufacturing costs, and product costs.
• Business performance and operational KPIs.

These data points help managers understand current business conditions and the relationships between different activities.

AI helps uncover trends and risks in business data

AI and advanced analytics can identify patterns that may be difficult to recognize by reviewing individual reports separately.

Depending on the available data, models, and implementation approach, businesses can use these technologies to:

• Forecast demand based on historical data and relevant influencing factors.
• Detect abnormal raw material consumption.
• Estimate the risk of production delays.
• Identify unusual patterns in equipment performance.
• Analyze factors that may affect costs and profitability.
• Support managers in evaluating operational alternatives.

An important distinction: not every analytical capability is AI, and not every ERP system includes all these functions out of the box.

The value comes from combining reliable data, appropriate analytical tools, and the ability to turn insights into action.

ERP helps businesses understand what is happening. Data analytics helps identify trends. AI can support forecasting and anomaly detection, giving managers more time to consider appropriate actions.

3. Five AI Applications That Can Create Value for Plastics & Packaging Manufacturers

Not every manufacturer needs to begin with a large-scale AI project. For CEOs, the more important question is: Which business problem is causing significant losses, and is the available data sufficient to address it?

Here are five applications worth considering.

3.1. Demand Forecasting and Production Planning Support

When customer demand changes, manufacturers must balance production capacity, inventory levels, and delivery commitments.

If planning relies primarily on experience or incomplete, outdated data, businesses may overproduce, run short of stock, or repeatedly adjust production schedules.

Forecasting models can analyze historical order data, seasonal patterns, and other relevant factors to help estimate future demand.

These insights can help manufacturers:

• Prepare more appropriate production plans.
• Estimate raw material requirements.
• Identify potential stock shortages or excess inventory.
• Communicate proactively with customers when delivery capacity is affected.

However, demand forecasting does not automatically translate into automated production scheduling.

To create a feasible production plan, manufacturers must also consider machine capacity, mold or product changeover times, labor availability, material availability, and other real-world constraints.

CEO perspective: Forecasting creates value when it helps the business prepare its resources more effectively, not simply when it produces a number representing future demand.

3.2. Detecting Anomalies in Material Consumption and Scrap

Raw materials often account for a significant proportion of manufacturing costs in the plastics and packaging industry. Even small, persistent deviations in consumption can therefore affect business performance.

For example, the actual material consumption of a product group may increase relative to its standard usage. Possible causes include raw material quality, operating parameters, specification changes, scrap generation, or inaccurate data recording.

If managers review only consolidated monthly reports, they may discover the problem after opportunities to address it have diminished.

When sufficiently detailed data is available, analytical tools can help compare actual consumption against standards, identify unusual trends, and flag cases requiring investigation.

For certain processes, computer vision can also support surface defect inspection or product classification. However, its effectiveness depends on the types of defects, operating conditions, image quality, and integration with quality control processes.

It is important to distinguish anomaly detection from root cause identification. Production teams still need to investigate and validate findings to determine the underlying cause.

CFO perspective: Instead of simply knowing how much scrap costs have increased, the business needs to understand where the increase originated, which products are affected, and where corrective action can reduce losses.

3.3. Identifying Risks of Production Disruptions

Equipment failure does more than generate repair costs. It can disrupt production schedules, increase waiting times, affect delivery commitments, and create unexpected expenses.

Under suitable conditions, analyzing sensor data, maintenance records, and operational information can help identify abnormal patterns before equipment experiences a serious failure.

This is the foundation of predictive maintenance.

However, implementation depends on whether the manufacturer has relevant equipment data, such as temperature, vibration, pressure, operating hours, or historical failure records.

For plants without equipment data collection systems, the first step may be to standardize maintenance records, document the causes of machine downtime, and track operating hours.

Manufacturers can then assess whether predictive models are likely to deliver sufficient business value.

COO perspective: The goal is not to predict every equipment failure. It is to reduce preventable disruptions and improve the ability to meet production plans.

3.4. Forecasting Costs and Identifying Profit Margin Risks

Increasing revenue does not necessarily mean increasing profits.

A large order may require more expensive raw materials, longer production times, higher scrap rates, or delivery arrangements that generate additional costs.

When financial and operational data are not sufficiently connected, businesses may struggle to assess the actual impact of these changes.

With the right data connected, analytical tools can help manufacturers:

• Compare actual costs against budgets and plans.
• Detect significant cost fluctuations.
• Analyze material consumption trends.
• Evaluate the impact of different raw material price or demand assumptions.
• Identify orders or product groups that require closer review.

For example, if raw material prices rise while selling prices remain unchanged, a manufacturer can analyze the potential impact on expected profit margins and evaluate options such as adjusting prices, optimizing material usage, or changing procurement plans.

AI can support this scenario analysis, but the results still depend on the accuracy of cost data, input assumptions, and calculation methods.

CFO perspective: The value of data is not simply reducing the time needed to produce a cost report by a few hours. It is identifying profitability risks before they become difficult-to-reverse financial outcomes.

3.5. Supporting Data-Driven Executive Decisions

CEOs do not need to review every machine parameter or raw material transaction. They do, however, need to understand which issues threaten business objectives.

Depending on the capabilities of the implemented solution, AI can help consolidate information, query data using natural language, or highlight significant changes.

For example, managers may want answers to questions such as:

• Which customer orders are at risk of late delivery?
• Which product groups are experiencing unusually high material consumption?
• Which factors are having the greatest impact on manufacturing costs?
• Which inventory items are at risk of exceeding forecast demand?
• Which departments need to prioritize corrective action this week?

In a suitably designed system, analytical tools can reduce the time spent searching for information and direct management attention toward issues requiring action.

However, the ability to answer these questions depends on source data, access permissions, integration, and the actual capabilities of the tools. Businesses should validate the accuracy of results, particularly when using generative AI to query or summarize business data.

CEO perspective: A good system does not replace the CEO’s judgment. It helps the CEO identify problems earlier, evaluate options faster, and spend more time on decisions that truly matter.

4. Why Implementing AI Before Data Is Ready Can Backfire

AI can produce highly convincing analyses. But a well-presented result is not necessarily an accurate one.

If input data is incomplete, inaccurate, or inconsistent, models may generate unreliable forecasts. In manufacturing, this can lead to inappropriate procurement decisions, flawed production plans, or ineffective resource allocation.

Before investing in AI, businesses should assess four factors.

Data quality

Are material usage standards updated when processes change? Are actual output, scrap, and machine downtime recorded completely? Does inventory data reflect physical stock accurately?

These questions may seem basic, but they directly affect analytical reliability.

Data connectivity

Can data from production, warehousing, procurement, and finance be reconciled?

If different systems use different product codes, classification methods, or recording rules, businesses need to resolve these inconsistencies before expecting a reliable, company-wide view.

Governance and processes

Who is responsible for validating alerts? Who decides what action to take? How will the business evaluate and adjust a model when its predictions are inaccurate?

AI creates operational value only when analytical results are incorporated into actual workflows.

Measurable business outcomes

Businesses need to define the project’s business objectives before implementation.

For example, if the objective is to reduce scrap, the company needs baseline data on scrap rates, associated costs, and the factors that may influence the results.

If the objective is to reduce production disruptions, the business needs to track machine downtime, incident frequency, and associated costs.

Without baseline measurements, it is difficult to determine whether AI has created meaningful value beyond adding another technology to the system.

The key principle: Do not begin by asking, “Where can we apply AI?” Start by asking, “Which business problem is most important to solve, and do we have the data to solve it?”

5. Where Should Manufacturers Start with AI?

Manufacturers do not need to deploy AI across the entire plant from day one.

For CEOs and executive teams, a more practical approach is to select a problem with a clear business impact, assess the available data, and conduct a controlled pilot.

A four-step roadmap can help.

Step 1: Identify the highest-priority business problem

Choose an issue that significantly affects costs, productivity, delivery performance, or profit margins.

Examples include increasing scrap rates, inaccurate demand forecasts, or excessive unplanned machine downtime.

Step 2: Assess existing data and systems

Determine where the required data resides, whether it is complete, how frequently it is updated, and whether it can be connected to ERP data.

If the data is not ready, prioritize standardizing data collection processes and improving data quality.

Step 3: Run a pilot with measurable objectives

Test the solution on a suitable production line, product group, or business process.

Define performance indicators before starting, such as scrap rates, forecast accuracy, machine downtime, or the time required to detect and address anomalies.

Step 4: Evaluate results before scaling up

Compare the results against baseline measurements, consider implementation and operating costs, and assess any changes needed to existing workflows.

Expand the solution only when there is evidence that it delivers benefits proportionate to its costs and risks.

This approach helps manufacturers avoid spreading investments too thinly and focus resources on applications with the potential to deliver tangible business value.

6. ERP with AI Capabilities: What Should Businesses Evaluate?

Not all ERP systems offer the same level of AI support.

Some solutions include specific analytics or automation capabilities, while other applications may require a separate data platform, analytical tools, or external systems.

When evaluating an ERP solution with AI capabilities, manufacturers should consider the following criteria.

Evaluation criterion	Questions for management
Data management	Does the system help standardize and connect the business data required?
Manufacturing fit	Does the solution support the plant’s specific processes, bills of materials, production planning, and management requirements?
Analytics capabilities	What reporting, analytics, or forecasting functions does the system provide?
AI integration	Which AI features are available out of the box, and which require additional implementation?
Implementation feasibility	What additional data, equipment, or infrastructure is needed to use these capabilities?
Governance and security	How is data accessed, used, and protected?
Return on investment	How will the business measure benefits and total cost of ownership?

How Epicor Kinetic fits into the evaluation

Epicor Kinetic is an ERP solution designed for manufacturers. It can be considered when evaluating an enterprise management platform and the potential to develop data, analytics, or AI applications.

However, businesses should verify individual capabilities against the relevant version, configuration, licensing, and implementation scope. They should not assume that every AI application is included by default or will operate effectively as soon as the ERP system goes live.

At Data V Tech Solutions, the evaluation process should begin with the manufacturer’s operational processes and business objectives. Only then should the business identify the appropriate ERP solution and supporting technologies.

A modern system does not automatically make a business more efficient. Its value depends on how it is designed, implemented, and used to solve real business problems.

7. What Should CEOs Ask Before Investing in AI?

Before approving a budget, CEOs should require their management teams to answer five questions.

The Competitive Advantage Is Not Having AI. It Is Acting Earlier.

In the plastics and packaging industry, manufacturers must continuously balance raw material costs, production capacity, quality, inventory, and delivery requirements.

ERP provides a data foundation for business management. Data analytics helps identify trends. AI can support anomaly detection, risk forecasting, and better-informed decision-making.

But for these capabilities to deliver tangible value, businesses need reliable data, clearly defined processes, and measurable business objectives.

For CEOs, the question is not whether the business needs to follow the AI trend. The real question is how many opportunities the business is losing by identifying problems too late, and whether existing data can help change that.

That is why ERP and AI should be considered as part of the same management strategy rather than as separate technology projects.

Explore How ERP and AI Can Support Plastics & Packaging Manufacturers

Data V Tech Solutions, in collaboration with Epicor, is hosting a webinar:

Breakthroughs in Plastics & Packaging Manufacturing with AI-Integrated ERP

During the webinar, participants will explore:

• Operational challenges that lead to losses in plastics and packaging manufacturing.
• Business problems that technology can help address.
• Practical approaches to using AI to extract insights from business data and support decision-making.

Date: Wednesday, November 11, 2026
Time: 1:30 PM Vietnam time (ICT, UTC+7)
Format: Online via Microsoft Teams
Language: Vietnamese
Speakers: Ms. Annie (Epicor) and Ms. Phuong (Data V Tech)

If your business is looking to improve operational control and make better use of data to support more proactive decisions, this webinar offers an opportunity to explore relevant approaches.

[Register for the Free Webinar]