AI is here – but first you need to get your data in order

AI has moved beyond hype to become a practical tool for manufacturers looking to improve efficiency, resilience and competitiveness. As adoption accelerates across businesses of all sizes, success will depend not on technology alone, but on high-quality data, strong governance and human oversight to ensure AI delivers real commercial value rather than costly mistakes. Shane Taylor, Manufacturing Sales Manager at ECI Solutions tells us more.

It’s been a year since the government announced its AI Opportunities Action Plan – a wide-ranging initiative that sets out how AI can be used to drive economic growth.

With major companies like Rolls Royce and JCB investing heavily in AI development, the technology is set to transform the manufacturing industry.

For manufacturers who have already become more data-driven, AI is the next step in their journey, a way to get more value from their data, rapidly and at scale. It should help them to become leaner, more innovative – and ultimately, more profitable and competitive.

The number of use-cases for AI in manufacturing is growing all the time. Smarter demand forecasting, inventory optimisation, dynamic pricing, accounts receivable risk prediction and automated document processing. It supports Internet of Things (IoT) technologies by capturing data and making fast decisions, for example, when to schedule machine maintenance to ensure top performance and avoid unexpected downtime.

After years of hype, AI is now a reality for many small and mid-sized businesses (SMBs).  Business software, like Microsoft, comes with AI tools, while many employees are using generative AI platforms like ChatGPT to help speed up daily tasks.

According to our own research, 93% agree that AI is critical to staying competitive and some have already implemented it. Around 28% now use it for marketing, 22% for customer service and 19% for data analysis and reporting.

It’s a promising sign that SMB manufacturers are embracing innovation like their bigger end customers. So, how can they be sure that their AI projects will deliver genuine value to their business and, importantly, don’t damage it with poor decision-making?

Data-first

The starting point for any AI project must be good quality data. It’s easy to upload Excel files to ChatGPT but you won’t get a reliable response if the information you input is old, limited, mistyped or recorded inconsistently by different users.

Yes, it’s the famous line ‘Garbage in, Garbage out’ (GIGO). When data is incorrect or incomplete, AI fills in the gaps or even hallucinates. Just one incorrect data point, such as quantity or price, can lead to poor and very costly forecasting.

Enterprise resource planning (ERP) software has already helped SMBs to become more data-driven and automated. It is also the foundation for AI-readiness – providing the comprehensive, up-to-date and standardised data needed to use AI tools effectively.

What do we mean by good quality data?

It’s all very well saying you need quality data but what does that actually mean? Widely-accepted frameworks, like ISO 8000, generally define it as accurate, complete, consistent, up-to-date, unique (no duplicates) and valid (conforms to defined rules/formats).

Any weakness across these areas means that AI model performance is reduced, leading to potentially inaccurate or misleading results.

You don’t need to be a statistical expert to achieve this. A modern ERP will do a lot of the leg work for you by making it easy for employees to record information in a standardised format. With data linked to orders, inventory and production, you’ll quickly build a complete and up-to-date dataset, which can be used to make AI decisions.

ERP vendors are also developing and embedding new AI tools and features in their software, allowing it to deploy it even if you don’t have a big IT team.

Simply by using the software, they naturally become part of an employee’s daily workflow. Well-governed, company-approved AI tools, delivered by an expert vendor and powered by reliable data, also reduces the chance that people will use other AI applications in a potentially risky way.

Last, but certainly not least, AI isn’t just a buzzword. It has to deliver commercial value. This is why, at ECI, our AI development is grounded in relevant and practical applications. What are the challenges manufacturers face today, and how could AI solve them?

Human oversight

No technology is infallible, no matter how well-designed it is. An ERP with AI capabilities promotes best practice but employees must understand the importance of inputting and managing data correctly.

Human experience and judgement is needed to confirm or challenge AI decisions, including recognising the current business and economic context and any ethical implications. They are responsible for leading on data governance, implementing and reinforcing policies, and ensuring that staff are fully trained.

Next steps

As we’ve seen, quality data is the starting point for AI projects. You first need to identify the areas where AI could deliver the most value to your business and then map out what data you need to achieve it. This includes identifying gaps, inconsistencies and potential risks. Dashboards in your ERP give you a clear understanding of what data you have and what’s missing, so you can ensure it meets the minimum standards for deploying AI.

Finally, remember that improving data is an ongoing exercise. Someone needs to own it but everyone is responsible for tackling the root causes of quality. Once again, this is much easier to achieve when you have an ERP that supports good data hygiene by design.

AI has become part of our everyday lives, and it’s an exciting prospect for manufacturers, especially younger people entering the industry who want to work with innovative digital technologies. Approved AI tools, powered by high-quality data, can help to future-proof your business, giving you the confidence to innovate and scale.