Embracing AI: understand the end user first

This is a guest blog post by Sara Portell, VP of User Experience and Rafael Artacho, Product Director, Unit4

How we interact with enterprise software, particularly enterprise resource planning (ERP) applications, is going to see its most radical evolution in the next decade thanks to artificial intelligence (AI). This is not hyperbole; it is the reality we will experience. However, the hard work starts now to convert the excitement around technologies like generative AI into a workable solution. There have been a lot of pronouncements about the integration of AI and what it will do for enterprise applications, but as with all new innovations we must be careful not to get too carried away. Technology history is littered with amazing ideas that failed to find a problem to solve. Likewise with generative AI technologies, like ChatGPT, we need to ask ourselves exactly what problems we are trying to solve before we launch into a full-on embrace of the technology.

So, what is the problem? Some of the historical frustrations with ERP systems include: it is clunky with manual and repetitive processes, unintuitive and information is siloed, slow and not geared to collaboration. Today, there is an opportunity to dramatically enhance the user experience thanks to AI and automation, but to be successful it is critical to understand the end user, what their business needs are and how the ERP system helps to resolve their problems. This is possible with the advent of frameworks like Design Thinking and technologies like microservices and cloud-based applications that ERP architects have been able to use to develop more responsive, intuitive systems.

We have been using research approaches from ethnography to deeply explore how users interact with ERP applications in the context of their work situation and IT environments. This has given us a detailed understanding of how they personalise their ERP experience by using shortcuts and workarounds, customising data input methods and using multiple devices. We have also discovered interesting trends in informal collaboration practices within the ERP environment and integrations with customized systems. In a variety of fields, much has been invested in delivering personalized experiences, in settings ranging from shopping malls to precision medicine in healthcare. The same principle is coming to ERP. What our ethnographic research is showing is that we can build up patterns of behavior which help us to understand the different use cases for ERP applications. From this, we are developing a framework for generative AI and a pattern library that will reflect the needs of individual users – for example, drop down menus will become far more tailored based on the user’s profile, previous interactions and typical responses; and there are many other opportunities to use generative AI in areas such as anomaly detection, smart recommendations and value prediction.

Ultimately, we want to create adaptive contextual experiences, where we follow a user-centric approach to understand the real problems facing customers and leverage AI to automate away tasks that are repetitive and mundane.  This will enable users to adjust their focus and better utilize their capabilities in more critical tasks. AI will help them to improve their decision-making by providing the right insights at the right time.

However, there is a ‘but’ which revolves around the same issue: understanding the end user. The reality is that AI will fundamentally change business processes. Our CEO, Mike Ettling, has talked about the potential efficiency gains that AI may offer by automating stages in a workflow. This will impact end users, which is why we are now rolling out a program to understand the change management implications of embracing AI. Job tasks are going to change which will affect roles, so at this stage we need to ensure all our staff understand and are evaluating AI in their context. Not just what it will mean for their careers, but how it can positively impact their workload, as well as understanding what they need to know as they put AI into practice. All these learnings will filter back into our strategy for integrating AI into our next generation ERP system, ERPx.

If users have not been exposed to AI in a work context and tasked with understanding how it may change their day-to-day tasks, there are a lot of considerations to evaluate before an AI application is released into a production environment. These include:

  • Data integrity: if the AI is using inaccurate data to make decisions it can expose an organization to bias or poor outcomes
  • Data privacy and security: it is critical to ensure critical information is not exposed to cyberattacks and there is no loss of control over data
  • Knowing the limits of AI: much has been made of what generative AI, in particular, can do in terms of rapidly trawling information to provide concise responses to complex questions, but it has limitations. If you are relying on it to build code that may end up in critical applications, do you know where that code has been sourced? Understanding the limitations of AI is key to a realistic approach to its adoption.
  • Interpreting the analysis: if you are being fed answers by AI how should you interpret them? Do you treat them as absolute truth or do you interrogate them? If you haven’t got a detailed knowledge of how AI decision making works this could be difficult, so how do you prevent employees becoming over reliant on AI generated answers?
  • Testing AI outcomes: a corollary to interpreting the analysis is ensuring there is an overseer who is in charge of verifying the decisions the AI is making and ensuring automated or AI-generated tasks make sense, are accurate and safe.

It is also critical to audit skills within your organization. The insights the Unit4 R&D team has collated underline the growing importance of human skills alongside technical AI proficiency. Equal importance should be placed on skills of interrogation and analysis as well as capabilities such as collaboration, decision-making, creativity and emotional intelligence as irreplaceable competencies to effectively leverage AI. Employees will also need a continuous change mindset and the flexibility to work alongside AI systems. These human-centric skills will become pivotal to empowering a workforce to innovate and strategize beyond AI’s capabilities.

AI and automation tools will continue to mature and become more mainstream in the years to come, so it is undoubted they will have a significant impact on the future direction of ERP development. What is critical is that we do not lose sight of the end user and ensure their requirements are central to any innovative use case for AI. Get this right and we will enter an exciting new period where ERP will work in tandem with users, who will develop a whole new range of skills and work opportunities.

Data Center
Data Management