For the past couple of years I’ve been telling my clients and community of business people that they have two choices when it comes to incorporating AI in their companies: build or buy.
Building, for the most part, has been a big brand game. As I’ve previously written, financial services companies from JP Morgan and other large firms have spent tens of millions of dollars building out internal platforms on AI models for customer service, research and programming functions. Smaller companies don’t have these kinds of resources, although more tools have come on the market that – with maybe a few hundred grand – can enable some to build out their own internal AI solutions. These tools will certainly improve and become less expensive over time.
Buying AI solutions, however, is a better option for most of these smaller organizations. But they have to wait until the companies that they rely on to provide their core software systems start rolling out AI features that are accurate and reliable enough to use. This is now happening, and the proof comes from one mid-market enterprise resource planning (ERP) provider, Epicor.
Epicor has been around for a number of years. The company serves more than 23,000 customers in 150 countries who are mostly distributors and manufacturers. It has grown through acquisition and by making a good product. Their recent user conference – where I participated (I have not been compensated by the company to write this column) – had about 4,000 attendees. Most of these attendees were accounting and management types eager to hear what’s coming from the provider of the core systems they use to process orders, manage inventory and make planning decisions.
They – like me – left with a number of takeaways. For me, these were the most significant.
AI in ERP is about business intelligence in 2025.
Epicor’s theme was “cognitive ERP” or using AI to analyze data and predict outcomes. By doing so, users can smooth out supply chain issues and potentially mitigate tariff challenges (a big topic this week) by better understanding their costs and product availability so that they can predict activity. A good manufacturing and distribution ERP systems enables companies to model sourcing scenarios and assess tariff impacts at the SKU or bill-of-material level, which is critical for navigating ongoing geopolitical uncertainty.
“You need to have a system that allows you to understand scenario planning and how you might source something differently,” Steve Murphy, the company’s CEO told me. “AI can help companies manage tariffs and supply chain disruptions by providing supplier risk scores, traceability, and “what-if” simulations-especially as geopolitical tensions rise.”
AI in ERP is also about supporting and not replacing humans.
The day will come when AI agents are reliable enough to truly do the work that humans are doing. But even ERP providers like Epicor aren’t stepping into that minefield yet.
“We’re not exaggerating. Summarizing specs, automating quotes, assisting orders…that’s all extremely real. But letting AI swap vendors? We’re not there yet,” said Murphy. “I wouldn’t fire anybody over AI.”
Vaibhav Vohra, Epicor’s President and Chief Product and Technology Officer, agrees and emphasizes that Epicor uses AI to automate routine tasks internally and increase margins to complement skilled labor by enabling them to focus on high-value work.
“Attempts to replace human workers with AI or robotics often fail,” he said. “We tried it and we weren’t successful. The true opportunity lies in using AI to assist workers and make them more efficient, not to eliminate their roles.”
Finally, AI in ERP will ultimately be about vertical integrations.
Vohra says that AI becomes significantly more useful when customized for specific industries. Which is why ERP software providers like Epicor are focusing on defined sectors.
“We’re moving away from being a general ERP provider to becoming a collection of highly specialized vertical ERP platforms, with each one tailored to its industry,” he said. “We’re picking our partners that we’ll be in bed with for a long time.”
AI without data is like a smartphone without the internet. But for most distributors and manufacturers, the data needed to make decisions isn’t just in their ERP systems. There’s data that will be collected by drones checking inventory, robots stocking shelves, machines that monitor and track runtimes and autonomous vehicles moving materials around a warehouse. Just as importantly there’s data in the systems of suppliers, customers, freight companies and other vendors.
Software companies like Epicor need to integrate their systems with all these other repositories to provide true value, and the executives running the company realize that. According to Vohra, the company has tripled its number of partners to focus on niche capabilities like rugged devices for poor Wi-Fi environments and other warehouse machinery.
“Our competitive strategy is rooted in specialization,” said Murphy. “By going deep into specific verticals like discrete manufacturing and hard goods retail, our company can differentiate itself from larger players trying to serve everyone. We’ll continue to be open to integrating or acquiring companies with complementary technologies like robotics or drones to deliver faster innovation to customers.”
All of this is where AI in ERP stands right now. In 2025. I expect this column to be outdated in a year, if not less.
“Despite all the complexity, our AI and data tools are developing more quickly than I expected six months ago,” said Murphy. “People are productive, and that’s working for us.”
Read the full article here
