When an exciting new technology emerges, companies often rush to test it—and then pause before making investments in full-scale rollouts. Some distributors are at that crossroads, suggests a new survey about AI adoption in distribution.

Of 233 distribution executives surveyed in December 2025 by Distribution Strategy Group, 63% said their companies were piloting or exploring AI applications, while only 27% said they already are implementing AI at scale. In other words, most distributors testing AI are still wrestling with whether to pull the trigger on rollouts.

“The industry has moved from the curiosity phase it’s been in for a year or year and a half to starting to execute,” says Brian Hopkins, chief operations officer of Distribution Strategy Group, which has conducted the State of AI in Distribution survey for the past two years. “Now we have to figure out how to operationalize it, so it shows up in actual performance.”

The good news for distributors not yet investing in AI is that most of their competitors also are in the early stages of AI deployment, so there’s still time to get into the game. But time is running out, Hopkins says. “Over the next two to three years, the ones first to test will be the ones who are able to win.”

The survey also suggests answers to some of the pressing questions being asked by executives just getting started with AI, such as:

• What parts of my business should I focus on first?
• Do I need to clean up all my data before starting with AI?
• What are the biggest obstacles to AI deployment?
• Who in the company should lead AI initiatives?

Hopkins and DSG co-founder Jonathan Bein emphasized on a recent webinar discussing the survey that, while distributors ponder these questions, vendors of ERP software and other technology are rapidly integrating AI into their systems. That can provide an easy entry point for AI adoption.

“You may not need an extra tool,” Hopkins says. “Talk to your enterprise resource planning (ERP) provider first. See what they have available.”

At the same time, Bein says it’s important to keep up with the fast-evolving AI tech offerings. For instance, he predicts the AI tool most often used by distributors today, ChatGPT, will soon be eclipsed by more capable rivals.

Growing Investment in AI Among Distributors

While there is hesitancy to make major investments, the 2025 survey shows many distributors see the value of AI. Among respondents, 67% of respondents said they expected their companies would increase their investment in AI in the next two years, while in the 2024 survey, in response to a slightly different question, only 43% said their firms were actively investing in AI.

That shift is particularly important because more C-suite executives responded to the 2025 survey versus the one in 2024 (68% of respondents in 2025 versus 53% the prior year). That suggests increasing executive focus on AI.

However, Hopkins and Bein say these surveys should not be taken as representative of the distribution industry because DSG surveyed companies on its mailing list who are more focused on technology than the typical distributor. Plus, the sample size was twice as large in 2025, which could mean less AI-mature companies participated in the recent survey.

For these reasons, Bein says, the results should be seen as indicating the direction the industry is moving and not necessarily providing an accurate snapshot of AI in distribution overall.

Where to Start with AI?

Companies contemplating moving from pilots to rollouts often wonder where the best place is to start. Some customer-facing functions are proving to be good options. For example, 62% of respondents whose companies are implementing AI use it in order automation, such as taking an order placed via email and automatically feeding it into the company’s ERP software.

Many customers of distributors generate orders in the form of PDFs or emails from their own ERP systems, and AI can accurately read those orders and feed them into the distributor’s ERP without an employee touching the order, cutting costs, Hopkins says.

Meanwhile, 41% use AI chatbots to answer customers’ questions. 32% are integrating it with their CRM systems, such as surfacing a customer’s order history and suggesting products they might need. It’s worth noting, however, that, given 68% have traditional CRM systems, many are not yet using AI in this way.

Deploying AI in supply chain functions appears more challenging. Only 28% said they are using AI in demand forecasting, 22% in inventory management and 10% in warehouse management or robotics.

Hopkins’ advice is to focus first on the areas likely to produce fast results. “Start with the easiest ones before getting to the harder things like demand forecasting and route optimization,” he says.

ChatGPT is the most often used AI tool by distributors, in use at 63% of those adopting AI. But Bein believes that it will change in the next period as competitors, notably Anthropic’s Claude chatbot, demonstrate greater functionality for business users.

One feature of Claude, which is more oriented to business users than ChatGPT, allows a company to take the approach of its most skilled employee in a certain task and create a “skill,” a bundle of instructions that everyone in the company can use easily.

Take how an experienced employee quotes an order, for example. “To produce a quote, I need to go to this piece of information, includes these things in the quote, have the right pricing, the right branding, all the terms and conditions built in,” Hopkins says. “All the things you would do to prepare a quote you build into a skill and that becomes the standard process for the company.”

Data and Employee Skills Remain AI Hurdles

Many executives worry that they must standardize their data before they can deploy AI effectively, and the survey suggests companies with more advanced data management are more confident of their AI success.

Respondents with modern, centralized data repositories are confident that their AI projects will yield ROI. But that’s true of only 18% of those who say their data remains siloed in separate databases.

But Hopkins emphasizes that data quality isn’t as important in some applications as it is in others. “You can have decent data on a specific AI but need very good data for other pieces,” he says.

For example, there are AI-based demand forecasting models that work well with pretty good data, Hopkins says. “But,” he says, “if you’re going to have robots in your warehouse the data better be good, because robots are moving around your warehouse and your people, and that’s where data has to be tied in well.”

While poor or incomplete data was the second most-cited obstacle to AI adoption in the survey, at 24%, people issues emerged as the biggest barrier. Between the 33% of respondents who cited a lack of internal skills as an AI roadblock and the 19% pointing to resistance to change, that’s more than half who believe people issues are the biggest impediment.

Hopkins says some of the resistance comes from veteran employees who believe they know the business better than the AI could, and doubt the suggestions that AI comes up with. Over time, he says, more employees will appreciate that AI can spot trends humans can’t, leading to better results and less resistance.

Notably, top management interest in AI showed up at the bottom of the list of AI obstacles in a separate survey question, an indication that it’s no longer hard to get the C-suite’s attention for AI proposals.

“Most leaders want AI,” he says. “The real constraint comes down to execution: the bandwidth of the organization, the skillsets of people, that’s what hinder execution.”

One way to address that execution issue, he says, is to run fewer pilots so that leaders can give them more attention. “People are running all these pilots,” he says. “Fewer well-done pilots are more likely to have success than a ton of them.”

Who Should Lead AI initiatives?

Among companies responding to the survey, 41% said executive leadership had responsibility for AI projects while 22% said it rested in the hands of technology or IT managers. Among the survey’s revealing findings was that projects led from the top are making more progress than the IT-led projects: Among companies scaling AI projects, 55% are led by executives and only 15% by IT teams.

The C-suite views AI as a business transformation tool, while IT is more likely to view AI as a technology project, the DSG executives say. IT-led projects also may bog down in lengthy data-cleansing efforts, which can slow down deployment, Bein says. That underscores the importance of top leadership being involved.

“We think AI is too important to be left to IT only,” Bein says. “Of course, IT plays an important role. But to the extent that IT is driving the whole thing you might see a data governance mentality outweighing some of the things we see with trying to get things done.”

At the same time, Hopkins says AI decisions should include front-line employees as well as top executives. He gave the example of a company he worked with where the executives proposed starting with an AI-based quote-automation system, only to hear from the people who do the quoting why that might not work. They chose another area to start.

That’s a lesson for other distributors. “Having the people closest to the work in on the decisions,” Hopkins says, “will help you make better decisions on what’s on the roadmap first.”