AI That Actually Works: Where Distributors Should Start and What Delivers Results

Distribution Strategy Group's Brian Hopkins shares five practical ways distributors can move beyond experimenting with AI and start generating measurable business value.

Artificial intelligence may dominate conversations about the future of distribution, but for many companies, one fundamental question remains: Where do we actually start?

According to Brian Hopkins, Chief Operations Officer at Distribution Strategy Group (DSG), the answer isn't necessarily an expensive enterprise-wide technology investment. It's finding everyday business problems that AI can solve, putting the right people in charge and measuring whether the results justify the effort.

During a recent episode of the American Supply Association's Embracing the Future podcast, hosted by ASA Director of Innovation & Technology Nils Swenson, Hopkins shared lessons from implementing AI within his own organization and helping distributors turn experimentation into operational improvements.

Start with the work you're already doing

When asked about getting started with AI implementation, Hopkins offered a straightforward example from his own daily routine.

Previously, he spent approximately an hour each morning reviewing emails, checking his calendar, researching upcoming meetings and catching up on industry developments. Today, an AI agent completes that preparation overnight, delivering a consolidated briefing he can review in 15 to 20 minutes.

He also described reducing an invoicing process from 15–20 minutes to roughly 90 seconds by teaching an AI tool to follow the company's existing procedure.

"It's using it to make the business more productive, so I can start to focus on other things," Hopkins explained.

Similar opportunities in distribution firms could exist in account research, order processing, quotation preparation and other repetitive administrative tasks.

Practical takeaway #1: Identify three recurring activities employees spend significant time completing. Document the current process, then evaluate whether AI can reduce the time required without sacrificing accuracy.

Turn repeatable tasks into consistent processes

One of Hopkins' most practical recommendations involves creating reusable AI workflows, often called skills.

Rather than repeatedly asking an AI model to perform the same task, Hopkins establishes instructions that allow the system to execute a defined process consistently.

"Anything I've asked AI to do for me more than three times becomes what they call a ‘skill.’"

For example, DSG has helped a distributor develop a library of AI skills that inside sales representatives can use to research unfamiliar customer accounts before making calls.

Hopkins made it clear that automation doesn't eliminate the need for oversight and instruction.

"You can't just build a skill and let it run," he cautioned. "You have to build the skill, but then you have to monitor the output."

Practical takeaway #2: Look for tasks involving repeated research, formatting or standardized procedures. Build reusable instructions, test the results and assign an employee to review outputs before incorporating them into daily operations.

Understand where the biggest returns may be

While order entry and quotation automation are common starting points, Hopkins believes another application offers even greater potential.

"The highest value is really the inventory optimization AI. And it's the highest value, but it's the least adopted."

AI-powered inventory optimization can help distributors make better purchasing and stocking decisions, potentially freeing working capital tied up in inventory.

However, Hopkins acknowledged that these systems can be more difficult to implement, making simpler applications a valuable starting point.

He also noted that order-entry automation can deliver substantial productivity improvements, citing potential gains of 30%–40%.

Practical takeaway #3: Separate quick operational wins from longer-term strategic opportunities. Start with manageable applications that build employee confidence, while evaluating higher-impact opportunities involving inventory and working capital.

Give AI the information it needs – safely

Product information, standard operating procedures, company policies and relevant customer context all influence the usefulness of AI-generated outputs. A common mistake, Hopkins explained, is expecting useful results without providing sufficient business context.

"It's the data. It's no different than any other software or system we've ever put in place."

But giving AI access to more information also introduces important considerations around data security and governance. Distributors need clear policies defining which AI tools employees can use, what company or customer information can be shared, and how sensitive data is protected. Without those guardrails, even well-intentioned AI adoption can create unnecessary security and privacy risks.

Hopkins also recommended organizing ongoing work into separate AI projects rather than relying on one continuous conversation for unrelated tasks, helping maintain relevant context and improve results.

Practical takeaway #4: Before abandoning an AI application because of poor results, evaluate whether it has access to the correct information, instructions and context.

Put someone in charge of the results

Perhaps the clearest distinction between distributors experimenting with AI and those successfully implementing it is accountability.

Hopkins' research found that companies moving beyond pilot projects generally had a designated business owner responsible for measurable performance.

"A lot of companies will automatically gravitate to this being an IT project. And it's not an IT project."

While IT plays an essential supporting role, Hopkins emphasized that operational leaders must define what success looks like.

"The real test is, is it hitting a measurable result for the company?"

Practical takeaway #5: Assign every AI initiative a business owner and a measurable objective, whether that's reducing invoice-processing time, accelerating quotations or improving inventory efficiency.

Find your internal AI champions

Hopkins offered one final recommendation to companies unsure where to begin their AI journey: Identify employees who are already experimenting with the technology.

"Find those people in your company. Make them the champions for your AI."

And the first project doesn't need to be revolutionary. "It doesn't have to be glamorous," Hopkins said.

Hopkins and Swenson agreed that successful AI adoption won't be measured by how many tools a company implements, but by the problems those tools solve and the value they create. The distributors that see the greatest returns will be those that empower their people to identify opportunities, turn ideas into action and hold every investment accountable for measurable results.

Listen to the full episode of ASA's Embracing the Future podcast featuring Brian Hopkins for additional examples, implementation lessons and practical guidance on getting started with AI.