Learn how to increase adoption and get AI working to streamline your ERP processes in 2025 and beyond.

 

In a Gartner survey from 2023 (quoted by Harvard Business Review), 79% of corporate strategists said that using AI, automation, and analytics would be critical to their success going forward. Yet only 20% said they are using AI in their day-to-day activities. This survey gets right to the crux of the issue with using AI these days: everyone agrees that AI is critical, but no one is using it. Why?

One reason is that people think it’s dangerous. The same HBR article also reported on research from Forbes Advisor, which uncovered that 80% of Americans think AI will allow criminals to use their personal data maliciously. They also quoted a YouGov poll that determined that almost half of all Americans believe AI will attack humanity in the future.

Another important reason uncovered by Harvard researchers is that people believe that AI is too opaque and complex to understand and too inhuman and rigid to be helpful. Since people perceive AI as a “black box,” they do not trust its results.

People have good reason for their fears. In general, as we’ve seen it deployed, AI is disappointing and often useless – and that holds true for the ways many ERP companies are using it too.

We believe AI used for mission-critical business tools like an ERP should be perfectly safe to use and reliably trustworthy. And its features should be strategy-centric, not just tech-centric, because the benefits of leveraging AI strategically are much more impressive than all those pointless bells and whistles that companies insist on putting in modern products (e.g., chatbots that provide unhelpful responses instead of actual tech support, or automated “smart” features that require constant corrections by the user, or “gamification” badges and achievement points in business or productivity software where they add no motivation.).

With that said, here is how you can strategically use an AI-enabled ERP in your organization and gain long-term value from it.

First: Overcome AI Worries and Fears

As we mentioned, the two main worries about AI revolve around safety and trustworthiness.

AI Safety

Since AI is based on the concept of machine learning, it requires massive datasets from which it can learn. Unfortunately, as many companies have found, any data you feed into a public LLM (large language model) will probably be used as a data training set for that LLM. And since AI generally remixes what it’s been taught into different outputs, that can mean your company’s confidential or sensitive data could be remixed and shared as output results for other companies.

The obvious solution to this issue is to withhold specific data from AI systems, which is what most companies are doing now. But we all know how ineffective data silos are. Plus, some ERPs have partnered with LLMs to roll out their AI features more quickly, so your data could end up as grist for the public LLM mill even if you are very careful about segmentation.

With AI in an ERP system, we firmly believe your data should be safe and yours. You should also maintain full governance over where and how your data is used. Make sure the ERP you choose bases all its AI decisions on foundational principles of data governance and ownership that you agree with.

AI Trustworthiness

According to Harvard researchers —and what we’ve all experienced— data “hallucination” is an extreme issue for lots of AI. The best way to curb these tendencies is to make sure that you check where the data comes from. With tools like ChatGPT, you can directly question the AI about its sources; with an ERP, you can make sure the AI allows you to drill down directly into its source information and provides instantaneously generated charts to help you visualize and interpret the raw data.

The experts at Harvard assure us that people naturally want to know why AI provides the answers it does. Drilling into raw data or viewing charts that elucidate an AI’s reasoning process is a great way to do this. Double-checking the AI’s conclusions could also help you and your team understand how to tweak a prompt better to get a more applicable answer, which is an invaluable skill to have as we venture into an AI-driven future.

Finally, checking on sources helps you train the AI. The game-changer in today’s AI, as opposed to prior computer programs, is that it is based on machine learning: in essence, you can adjust its behavior and responses over time (and with patience). Suddenly, we have all become teachers of AI, which is a willing but foolish student.

Next: Get Your Staff to Embrace AI

Again, HBR offers some great advice to help you get your team on board with AI:

• Do a slow rollout. Too many AI tools at once can lead to analysis paralysis.
• Start with simpler models first. Simple models are easier to understand, as well as easier to catch in an error. Don’t make the mistake of thinking a “simple” model is useless, though: the elementary Turing computer broke the Enigma code in WWII. “Simple” chatbots can powerfully transform data into an understandable narrative and, of course, like the Turing computer (but much better), can crunch data at an astonishing speed. Plus, “agentic” AI is based on stacking multiple, simple AI functions into a workflow so the AI can complete very complex tasks.
• Ensure your team feels in control of the AI. Did you know that iRobot programs the Roomba to follow predictable routes across your floor? They found that if the Roomba moves unpredictably, people get the unpleasant feeling that it’s “alive.” To help your team feel that your AI is predictable, you may want to have it double-check report parameters with humans before executing a report or otherwise allow for on-the-fly tweaking from human operators.

Focus on the Best Uses for AI in an ERP

When we said that AI should be “strategy-centric instead of tech-centric,” we meant that AI usage should always begin with what the user needs rather than what the AI can do.

We’ve all heard that ChatGPT can write instructions for removing a peanut butter sandwich from a VCR in the style of the King James Version of the Bible (see the instructions on X / Twitter) – but can you imagine any possible universein which you would need your business ERP’s AI to do that? Neither can we.

By this point, we have all used software with pointless AI features that are awkward bolt-on capabilities. Instead of allowing yourself to be swayed by such “gee-whiz” features, we recommend that you think of AI as a force multiplier that reduces wasted effort and time as your company uses it to do what you always do—but better and faster.

Here are 2 key ways you may want to consider using AI right now:

• Data Translators

AI is known for its ability to crunch a staggering amount of numbers at an incredible speed. Combine this capability with the power of natural language processing (NLP) chatbots, and you’ll instantly gain the power to make data accessible to everyone in your organization securely.

For example, AI could answer a question like, “If we shut down this one specific plant for the next 3 weeks for upgrades, will we still be able to meet the current demand?”

Without AI, your staff could be required to run and analyze dozens of reports to answer that – yet AI can crunch the numbers, as well as analyze and interpret the data for you, in seconds – and then give you a clear, quick, yes or no answer.

• Intelligent Advisors

We mentioned “agentic AI” earlier. This concept, which you’ve probably run across in the news, means that the AI can perform tasks with or without human interaction. Researchers from Accenture tried a service called DoNotPay, an agentic AI that can identify opportunities for consumers to save money on recurring bills or parking tickets – and then automatically draft letters and emails to reduce those charges or cancel unused subscriptions.

The key here is that the AI acts as an agent on the user’s behalf to accomplish a task with little to no handholding – like a great employee.

Intelligent AI advisors could act as agents to scan your ERP for compatibility issues with upcoming updates; schedule updates to run when the system is not being fully utilized (non-peak hours); make recommendations and changes to optimize your complex workflows; or take care of sequential, rote tasks for you quickly.

Case Study: Exemplary AI in Acumatica Cloud ERP

To sum up everything we’ve said, AI features in an ERP should be in the long-term best interest of the ERP’s customers—not just a flashy, poorly thought-out PR ploy. That’s why we’re excited about Acumatica Cloud ERP’s rollout of AI capabilities.

They have:

• Secure AI

To Acumatica, security is not just a feature – it’s core to the system. Their AI keeps your data fully compliant, encrypted, and isolated from public LLMs, meaning your Acumatica data will never be in a public LLM.

• Practical AI

The Acumatica team has chosen to approach AI in a practical, strategy-centric way. Instead of making their ERP distracting and harder to use with new AI features, they rolled out their initial AI capabilities (anomaly detection), paired with a significant update to the user interface, so their AI is easier to use along with everything else.

They are committed to delivering, as they put it, “AI that just works for you.” In other words, it is the force multiplier that helps you do what you usually do, but far better and far faster.

• Their AI chatbot delivers clear answers with charts and drilldowns, without you having to perform clicks, queries, or reporting. You just ask a question, and you get a trustworthy answer.
• Their AI workflows allow users to connect multiple AI-driven functions to achieve agentic AI. For example, the AI can identify a supplier delay and then automatically draft a notification email for affected customers.
• Their AI optimization constantly monitors your system and makes proactive recommendations on how to improve ERP performance. It also checks for compatibility issues with customizations before running an update, helping you ensure everything goes smoothly.

In addition, Acumatica cleanly integrates with third-party AI solutions to help you use the AI tools you already like while maintaining proper data safety and accuracy on your behalf. It also has an AI development tool that empowers you to build your own AI capabilities if you’d prefer to do that.

How to Make Sure You Get Long-Term Value from Your AI-Enabled ERP

It may seem uncanny that Acumatica delivers all the hallmarks of practical, long-term ERP, but it’s really just the result of careful strategic planning. As they have repeatedly stated in interviews, they plan 3 to 5 years into the future, so they provide not only the ERP their customers need now but also the ERP they’ll need a few years from now.

One way that Acumatica plans ahead is with the Acumatica xRP, the extensive platform that Acumatica Cloud ERP is built on, which supports blazing performance, complex workflows, critical automations, and nearly infinite integrations.

But, again, the xRP is only one of the many ways Acumatica focuses on being flexible for the future and built to power your business growth.

When choosing AI capabilities for your ERP to provide long-term value for your business, it pays to work with an ERP provider that is genuinely committed to delivering long-term value for their customers.

 

Access the Solution Brief and Discover the Guidelines for Selecting an ERP Platform That Will Deliver Long-Term Value

 

Accounting Business Solutions is a respected accounting and business management software solution provider. We provide experienced sales, consulting, implementation, training and support services for small to medium-sized companies located in south central and southeast Texas. Our portfolio of solutions includes cloud-based, hosted, and on-premise options.

 

­­­­_________________

Sources

Harvard Business Review, “Why People Resist Embracing AI,” January 2025. https://hbr.org/2025/01/why-people-resist-embracing-ai

Harvard Business Review, “How Generative AI Improves Supply Chain Management,” January 2025. https://hbr.org/2025/01/how-generative-ai-improves-supply-chain-management

Harvard Business Review, “The Secret to Successful AI-Driven Process Redesign,” January 2025. https://hbr.org/2025/01/the-secret-to-successful-ai-driven-process-redesign

X (Twitter), “Write a biblical verse in the style of the king james bible explaining how to remove a peanut butter sandwich from a VCR,” December 2022. https://x.com/tqbf/status/1598513757805858820?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1598513757805858820%7Ctwgr%5Ee7a50ca43a9515c281d756cd38a3dbf45adfb589%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Fwww.nytimes.com%2F2022%2F12%2F05%2Ftechnology%2Fchatgpt-ai-twitter.html

IBM, “What are large language models (LLMs)?” November 2023. https://www.ibm.com/think/topics/large-language-models