In a previous blog AI and Machine Learning in Manufacturing ERP: Key Benefits, we discussed the benefits of using AI in manufacturing and how it could be enhanced with an ERP system. While manufacturers are keenly interested in using AI, the main question they have is what are the best use cases for AI in ERP?
There are so many options for using AI in ERP that it can be tempting to rush into a project without thinking it through properly. Despite the hype around AI, many manufacturers have still to implement it in their manufacturing systems. According to SYSPRO research with Frost and Sullivan Complex Manufacturing Outlook: Emerging Trends and Opportunities, only 19 percent of companies in the complex manufacturing sector have adopted AI. In the mid-marketing manufacturing sector, the adoption is lower.
Where AI can add value to ERP
As was pointed out in the previous blog, there are many areas where AI can benefit a manufacturing ERP.
- Using AI to predict when maintenance will be required to prevent unexpected downtime and to extend equipment lifespan.
- AI can be used to improve demand forecasting, enabling better inventory management.
- Vision systems incorporating AI can inspect products for defects in real time, ensuring consistent quality and reducing human error.
- AI algorithms can analyze production data to optimize schedules and allocation of resources, increasing throughput and reducing production costs.
- AI can provide real-time insights and analytics, enabling manufacturers to make informed decisions based on accurate data.
- For product design, generative AI can analyze large datasets and existing designs to come up with new design concepts that would be unlikely with traditional methods.
A new use case of AI in ERP is Agentic AI, which refers to AI designed to operate autonomously and make decisions on behalf of humans or other systems. Agentic AI can adapt and learn from past data, allowing it to perform complex tasks without constant human supervision. It provides the potential for virtual assistants which can understand and process natural language more effectively.
Challenges of using AI in ERP
One reason why AI initiatives have not been adopted more quickly is the experience manufacturers had with the high costs associated with Industry 4.0 deployments, which did not often provide short-term benefits.
Manufacturers who implemented Industry 4.0 or Smart Manufacturing in the past can see many of the same problems with AI:
- fragmented or poor-quality data,
- integration with existing systems,
- workforce readiness.
AI is only as good as its data
The right data in the right format is the foundation of AI success. Companies that fail to check and provide the right business data for AI will find that AI produces unreliable results. The adage of ‘garbage in, garbage out’ still applies with AI and manufacturers that want to start an AI project must invest in cleaning their data first.
AI requires a unified, consistent dataset, but many companies operate with multiple, disconnected systems that contain overlapping but inconsistent data. These need to be consolidated before implementing AI to avoid misleading predictions.
Since AI became so prominent in 2023, there has been a perception that AI requires large language models (LLMs) and that the bigger a model, the better it is. Only a few companies can afford the infrastructure and other costs of this type of AI. However, small language models (SLMs) are now appearing that offer significantly improved costs, and more importantly, the fine-tuned and specialized knowledge that manufacturers will require.
Integrating AI
For any AI system to be useful, it must be able to integrate with data in other systems. The advantage of AI as part of an ERP is that the integration is built in, therefore the data is already available for the AI system to use.
Workforce not ready for AI
Many manufacturers are struggling to recruit and maintain sufficient staff just to implement ordinary operational improvements. Very few companies have the level of maturity – skill, knowledge, understand – to make AI initiatives succeed. Since many staff are apprehensive and need support, the transformation needed will take time.
How to start an AI project
Several experts in the ERP arena have the following advice.
- Identify a key area, for example, demand forecasting in the supply chain process, which can benefit from AI. This will help to create momentum and build trust across the organization.
- Focus on integration with other systems: this ensures that new technology complements existing systems rather than replacing them outright.
- Ensure data collection and management is prioritized. It is critical to gather accurate and relevant data from various sources within the manufacturing process.
- Select the right AI solution that aligns with business needs.
- Start with a pilot project to test the effectiveness of the focused AI project. This makes it easier to evaluate the benefits and make adjustments before full-scale implementation.
- The human factor in complex projects is crucial. Therefore, get buy-in from staff on the AI project, involve them in the implementation, and train them on how to use the AI tools.
The potential of AI in ERP
The integration of AI in ERP can help to revolutionize how manufacturers operate. However, despite its promise, the adoption of AI remains low. To harness AI effectively, manufacturers must approach implementation with due care and without rushing into it. By following a structured roadmap, manufacturers can unlock the potential of AI in ERP, driving efficiency, innovation, and competitiveness in an increasingly complex market.
