With AI-native startups flooding the market, there is speculation in the industry that enterprise resource planning (ERP) systems will become entirely obsolete. The assumption is that generative and agentic AI systems have advanced far enough to help companies move away from complex enterprise platforms that have long dominated business operations. However, this is not entirely true.
The truth is that companies will never be able to fully eliminate the role of ERPs in their operations. They’re the backbone for maintaining critical business functions, such as compliance, auditing, governance, and costing, and serve as significant systems of record.
That said, one can’t deny that AI technologies are transforming the way companies interact with traditional ERPs. Specifically, AI is adding a layer of automation – compared to the manual ERP interactions of the past – to reduce the complexity and friction that come with ERP use. Research by McKinsey reveals that AI agents can make ERP implementation 50% easier and even accelerate its program duration.
For example, here's how an automation-led ERP might work in a manufacturing setup. A floor supervisor, who previously had to pull up multiple ERP reports to analyze the status of production, can now rely on an AI assistant for metrics like downtime or production line output. AI extracts data from ERPs in seconds using machine logs or shift reports to summarize production status or the causes of downtime.
However, there is another facet to this process. As AI integrates into enterprise workflows, it is highlighting a long-standing, hidden ERP issue that companies have yet to effectively address: data quality. The quality of data available at the outset can influence the analysis AI systems generate. If companies feed bad data into ERPs, any AI-powered analysis from it becomes suspect at best. About 69% of companies report that bad data hampers the decisions and insights their AI systems produce.
In this article, we’ll discuss how AI systems are modernizing traditional ERPs. And how, in doing so, they bring to the fore the importance of data quality and how it can affect the way AI-fueled ERP functions.
Enterprise AI Needs Clean Data to Function
When AI accelerates ERP workflows, that speed comes at a cost, especially if your operations are riddled with poor, unreliable data. Therefore, while enterprise AI delivers decisions and insights at great speed, it also amplifies bad data faster. If this kind of data becomes widespread, the very technologies used to harness it may become ineffective.
In an interview with MIT Technology Review, Irfan Khan of SAP Data & Analytics confirms this assumption. He says, “AI is incredibly good at producing results. It moves fast, but without context, it can't exercise good judgment, and good judgment is what creates a return on investment for the business. Speed without judgment doesn't help. It can actually hurt us.”
In contrast, if the underlying operational data is timely and correct, this kind of enterprise AI acceleration can be powerful. It can significantly reduce manual effort across teams, helping companies respond to crises faster and make better decisions.
There’s another reason why bad data can upend even the strongest enterprise AI systems. By design, AI models accept ERP data as gospel. They believe that the data reflects the reality on the ground, whether it’s output counts on the manufacturing floor or medication records in a healthcare setting. This can have disastrous results. For example, if an employee records inventory incorrectly in the ERP system, it can cause AI systems to make the wrong purchasing recommendation, further affecting downstream processes.
Since ERPs are purely record-keeping systems, they, like AI, do not question the data they receive, which can lead to problematic insights down the line. As such, what may have once been a localized data issue instantly becomes an enterprise-wide decision problem. Gartner predicted that poor data governance was one of the major reasons why companies would abandon 30% of generative AI projects.
ERPs Are Important, But They Cannot Fix the Data Problem at the Source
Companies that use AI to modernize their ERPs without prioritizing data quality overlook one major caveat. ERP systems, while invaluable, are not systems of capture. They only serve to maintain and structure business data and operational records. They also manage workflows and enforce rules that ensure auditability and compliance. But here’s what they can’t do: control how companies collect data in the first place.
In many companies, data is scattered and fragmented, with technicians collecting information on paper forms in the field or inspectors logging quality results on spreadsheets.
ERPs may only see this data hours or days after the actual work has happened. Even then, it may have been altered due to manual entry errors. As a result, ERPs receive bad data to begin with, and even AI cannot salvage it. Therefore, the focus has to shift from fixing the ERP to fixing the process of capturing data.
The Good News: AI Is Exposing the Hidden Problems in ERPs
In addition to reducing the friction that comes with using legacy ERP systems, AI is inadvertently shedding light on data quality issues that previously went unnoticed. Before the dawn of automation, companies dealt with imperfect data through manual intervention. Managers added context to reports, while seasoned staff filled in gaps in spreadsheets using their judgment. This kind of corrective layer temporarily fixed data issues, but it never addressed the root cause.
However, when AI stepped in, the entire dynamic shifted. In a real-world setting, AI systems skipped the step of “second-guessing” and simply analyzed the available poor data from ERPs. As a result, systems produced and amplified flawed analyses and recommendations. As AI brings such system-level errors into sharper focus, companies can no longer ignore looming issues in operational data. Over time, they are likely to snowball into business risks that can threaten the bottom line.
The good news is that AI is changing the way companies approach their enterprise systems. More companies are now forced to acknowledge the shortcomings of ERPs and the inability of AI to produce insightful outcomes without clean data. It highlights why companies shouldn’t wait to roll out ERPs before fixing their data first.
Why Point-of-Work Data Capture Is Crucial
When it comes to ERP and AI adoption, one thing is becoming increasingly clear: companies need to fix the first mile of data instead of waiting for tools to analyze or correct it later. This means capturing data where the work actually happens in the real world. For instance, discrepancies in inventory are common in a warehouse environment.
But why do they happen? Because staff often delay reporting stock fluctuations and movements as they occur. They may scribble entries into unofficial notes, only to enter them later into an unsynced spreadsheet.
However, if employees recorded crucial stock movements at the time of occurrence, either by scanning a barcode or via a mobile app, the inventory would remain accurately updated in ERP systems. Any restocking decisions or purchasing recommendations made by AI would then be based on actual scenarios rather than conjecture.
This is just one example of how point-of-work data collection can lead to successful AI outcomes. Accurate and timely data capture can lead to significant benefits for companies, regardless of industry.
This is exactly the kind of problem Alpha Software specializes in solving. It fixes the root cause – data issues – that can confound both ERP and AI adoption. Alpha Software uses low-code applications to help companies build bespoke apps that allow service teams and frontline workers to log data on the go. The data is instantly synced with ERPs, often even in the absence of connectivity.
Ensure your ERP and AI systems work in your favor. Alpha Software helps you build better forecasting and decision-making systems by fixing your data.
Book a one-on-one expert consultation at Alpha Software today to learn more.
Comment