Don’t get too taken in by the hype around AI in 2025 – – Enterprise Times

This post was originally published on this site.

Enterprise Times recently received some predictions for 2025 from Claus Jepsen, Chief Product and Technology Officer, Unit4. Enterprise Times discussed some of these with him in an interview. There is a lot of hype around AI, and as 2025 unfolds, everyone is talking about Agentic AI. He noted, “There’s been a lot of hype around these agents this year, but we’re going to have to monitor how this technology comes to fruition in the next year to see if the hype matches the reality.”

Agentic AI is an evolution rather than a revolution, and is often perceived as the third wave of AI. ERP has always been about automating processes, and Agentic AI is just another tool to help increase the intrinsic automation within ERP.

A change occurred when ChatGPT was released. There is an acceptance that the new technology is usable. He explained, “We are seeing greater appetite among our customers to start using and applying these technologies into the ERP, in conjunction with the ERP workflows and processes.”

Unit4 uses the Microsoft framework for AI agents and is capable of having chatbots talking to chatbots. However, it must be driven by user requirements. Jepsen views this as an orchestration piece. However, it is more important to focus on what customers want. He added, “We are more focused on how we make the processes work more fluently for the end user’s perspective.”

The importance of the user experience

The days of functionality as a differentiator are ending. Jepsen predicted, ”It will not be the ‘what’ that differentiates ERP solutions but the ‘how’ you can use them.”

AI will only exacerbate this shift. We are heading to a point where people don’t log in to an ERP. The ERP will analyse whether it needs assistance and meet you where you are. That might mean within Teams, Zoom, Email, or Slack, for example.

It moves the ERP from a solution that requires data maintenance to one in 2025 that will become a different user experience altogether. One where the basic user experience becomes a fully automated system. You won’t have to interact with the ERP itself.

What is Unit4 doing? Jepsen said, “What we are doing as a vendor is embedding those tech capabilities. I see it as the best way to move forward in this next generation of interaction with ERP software.”

Is AI the answer for UX?

In his predictions, Jepsen was clear that GenAI is not the answer to everything. He commented, “We should be careful not to confuse the hype around conversational AI with AI for ERP systems: GenAI cannot solve all the requirements in the world of ERP.

“Conversational AI tools can be very open-ended in terms of the prompts they answer. They are also good at creating summarized insights based on all the information they can scan. However, to predict what the user will ask and, therefore, the information that the AI needs to provide is challenging. Precisely because the questions are so open-ended.”

AI often struggles to understand the user’s intent. GenAI can understand the context of what the user asks and retrieve information. However, you cannot just throw AI at an ERP system and expect it to solve all your challenges.

There are three important components that need to come together in the future. Jepsen explained, “You need to feed information into the system, you still need to plot out a path, and you need to map the intent of the user to something concrete.”

AI+ERP, not an oracle

This means an AI+ERP solution is not all-knowing and cannot run a business. Currently, no AI knows everything, and it may not be able to answer every question that a user asks. As a vendor, does that mean you are capturing each of those omissions to improve the AI? It is one of the advantages of the cloud. A Cloud-based ERP solution can collect all the available information to improve the offering. Is Unit4 doing this today, though?

Jepsen answered, “What we are focused on right now is to take the first step towards this world of revisiting the usage of the ERP. Using these technologies and changing it from a mindset of having to enter the system instead of running everything outside the system.

“We focus on building a better experience around, for example, inventory management, statement processing, matching, the financial close and other core processes. We’re also developing capabilities around how you contextually help the user if they get stuck. We don’t have this feedback mechanism now, but we should.”

On Security

In his predictions, Jepsen stated, “New regulation, such as NIS2 and DORA, will mean organizations must put more controls in place around data access and scrutiny of the ways vendors manage this process. This will create challenges around balancing investment in cybersecurity versus other priorities.”

It is the vendor’s responsibility to keep up with compliance regulations. Verticals such as finance, banking, insurance companies, and the public sector are also under increased attention and scrutiny.

However, whilst the vendor will protect the data it holds, where organisations build integrations to third-party products, they need to ensure that they have the same considerations around security, privacy and sovereignty.

Another area of concern is data privacy. The challenge for both vendors and customers is it is getting increasingly complex. An example is C5 (Cloud Computing Compliance Criteria Catalogue), a relatively new regulation that deviates from ISO 20001. Vendors and organisations will have to improve security parameters regularly to meet this new raft of legislation. Unit4 has already done so.

In addition, the complexity of PII, data sovereignty and privacy is becoming clear. Global organisations have to understand who is allowed access to PII data when employees and customers are working in different regions across the UK, EEC, EU and elsewhere.

On Sovereignty

Where data resides is also seeing increased focus from organisations and regulators. Organizations need to address data sovereignty to comply with national and regional legislation. In the worst-case scenario, particularly for mid-sized organizations that do not have the resources of larger competitors, they may have to evaluate whether it is cost-effective to enter certain markets given the regulatory requirements they will face.

I asked Jepsen how this would be solved and if data would become decentralised. While that may happen, Jepsen sees tools emerging, especially in insurance. He cited a product called CyberArk, adding, “What that allows you to do is to grant just-in-time access to data to a certain element of your estate. The customer can issue access to vendor employees for that data. That may solve this problem for a period of time because the customer then controls the access through this tooling.”

The next step is having an edge architecture that uses a distributed model with a central core processing engine. A multinational company will have different rules depending on where they operate. However, with over 220 countries worldwide, this might become impossible to navigate.

AI and Jobs

There is a debate about whether AI will supplement, remove or create jobs. What will happen as organizations automate systems?

Jepsen replied, “We believe that by removing some of this workload through automation, we can free up time for our customers’ employees. They can then do something that creates more value for their customers and focus on other things that generate more value for their company. Of course, we can’t decide what a CFO decides to do.”

He argued that you could see software engineers’ whole existence and role as making people jobless. However, automation often leads to the creation of new roles. Notably, automation during the Industrial Revolution led to the loss of rural jobs and the creation of factory jobs in cities.

Rather than doing data maintenance, you need people to look at and act on the insights. The better the data you obtain, the better the insights. While the new roles will require new skills, people can be trained.

2025 is halfway through the decade

Looking back, Jepsen predicts, “The first half of the decade has been incredibly challenging. But it has shown how and why moving to the Cloud offers every organization the ability to adapt to changing environments and engage with their customers via digital apps and services.

“We will continue to see more and more organizations of all sizes move to the Cloud, but it will be incumbent on vendors to ensure the path to the Cloud is as straightforward as possible and that they support customers on an ongoing basis to maximize the value of their investments.”

We have now reached a tipping point for the adoption of cloud-based software. Many organizations have already moved significant portions of their operations to the Cloud. They have recognized the benefits in terms of security, scalability, flexibility, and cost-efficiency. Jepsen also pointed out that this shift enables businesses to respond more swiftly to market changes and customer demands.

While the tipping point for the cloud has been reached, we are not yet seeing a shift where organisations are switching between one cloud ERP vendor and another. With the lifecycle of an ERP around 8-12 years, it will still be a few years before systems are replaced. One of the selling points that some cloud vendors used when first selling their solutions was the ease of switching to another vendor. The reality is different.

The challenge of migrating is still around the data. Jepsen sees a future where ERPs have a metadata-driven setup, and there are common processes between organisations. That could also lead to more fluidity between solutions. In that case, onboarding could become faster.

Jepsen concluded, “You will end up where ERP and a lot of enterprise software become these background processes you don’t really deal with, and you don’t really care, as long as you can get your finances out or timesheets in or whatever you need to do.”