AI pays off when you redesign how the work gets done

Benj: I think we will look back 10 years from now and think, "I can't believe that we had human beings doing some of the work that today is manual." We'll be spending the same dollars, but doing activities that make the experience for our customers way, way, way better. We'll think, "I can't believe we used to spend money on tasks that didn't actually matter for our end customer."

Hi, I'm Benj, founder and CEO at Proton AI, the AI-powered CRM built specifically for distributors. You're listening to In the Mind of a Distributor, where we interview the smartest minds in distribution. Let's get into it.

Dasha: Hi, you're listening to In the Mind of a Distributor. Today we're doing things a little differently. Instead of Benj interviewing a guest, I'll be interviewing Benj.

I'm Dasha, VP of marketing here at Proton. I've worked closely with Benj for four years. We're going to talk about his life story, how AI is changing distribution, what will stay the same, and what will look different. I hope you enjoy our conversation as much as I did.

What's up, Benj? How's it going?

Benj: It's good. How are you?

Dasha: I'm good. How the tables have turned.

Benj: I would much rather be interviewing someone than being interviewed.

Dasha: I can tell how uncomfortable you are, and it's kind of making me happy. For people who don't know your story, you grew up in your family's distribution business. What was that like?

Growing up on the customer's ground

Benj: We sell dental products to dentists all across the U.S. Growing up in the family business mostly looked like getting in the car with my dad and going co-traveling. He really believes that in distribution, the magic happens on the customer's ground.

He and the rest of the executive team spend a lot of time visiting customers and traveling with sellers. We'd get in the car with a rep and visit eight to 10 dental offices in a day. We'd watch how the seller interacted, learn from the customer, and meet the customer.

The other thing we spent a lot of time doing was visiting vendors. As a distributor, the other half of the business is having great relationships with vendors.

The most exciting vendor visit was a family trip to Korea that was really a secret way to meet Vatech, which has become a leader in dental X-rays. At the time, they were just getting into the U.S. market. I think we were their first distributor.

I was about 12, and I remember meeting Vatech's CEO. He described how they weren't the number one X-ray company yet, but they were going to be. Ten or 12 years later, they are a leader in the U.S. market.

Dasha: That's inspiring. You watched this guy predict, "We're going to be the best," and now it's true.

Benj: It was so inspiring. I remember him saying, "We will be the number one player in this market." He called his shot, and they've executed incredibly well.

They've also become one of the largest suppliers to my family business. That visit mattered for building the relationship. They're a family business, we're a family business, and we got to meet on their turf in Korea. It was awesome.

Dasha: Was your dad coaching you before these trips? You're 12. Is he telling you who you're going to meet and what you need to say?

Benj: No. He was like, "Just ask whatever questions you have." I've always been a crazy curious person. Growing up, before I asked a question, I would always say, "I have a question." My family would get angry and say, "Just ask the damn question."

On these visits, I was always asking 10,000 questions. There was no coaching. I think at that dinner my sister, who was about five, was so jet-lagged that she fell asleep on the table. I remember my dad wondering, "Is this okay? I don't really know."

The summer that led to Proton

Dasha: Then you went on to study applied math.

Benj: That came from my experience in the family business. In high school, I spent my summers rowing for the junior national team. Rowing is an amazing sport, and also a little boring if that's all you're doing. You wake up, row, eat, sleep, row again, eat again, and go back to sleep.

One summer, while we were training for worlds, I asked my dad if there was any work at Benco I could do remotely. Mark Kolanoski was leading the data analytics function within marketing, and my dad told me to call him.

Mark said, "We've got this big data set. We're trying to predict customer churn." He helped me download RStudio, and I taught myself R with his help. He would give me problems and mentor me. He tragically passed away a couple of years ago, but he was the reason I got into applied math in the first place.

It was this semi-fake internship where I built basic models for my family business while rowing over the summer. I loved math in high school, but that summer I understood what I found satisfying about it. I could follow my curiosity, explore the data, and deploy something that someone would actually use. When I got to college, I thought, "I want to do more of that thing."

Dasha: That's also the early story of Proton.

Benj: Exactly. I did some research at Harvard Business School one summer. The next summer I worked at Danaher, one of the manufacturers that supplied my family business. My plan was to work at Benco after college, so I thought I should spend some time on the supplier side and learn more about the dental market.

At Danaher, I was doing data analytics when the marketing team had an idea: maybe we could use the data to figure out which customers would want to buy certain products. I took that work and did similar consulting for other manufacturers.

I was talking to my dad about it, and he said, "You should do this for distributors. This is way more interesting than doing it for a manufacturer." That was the beginning of the company.

From PRM Solutions to Proton AI

Dasha: What did you call the company first? You should tell the story.

Benj: The original idea was to use data to predict how you should manage customer relationships. Instead of a CRM, it would be a PRM: predictive relationship management.

I bought prmsolutions.com for $5,000, which I still own and would love to sell to someone. At the time, $5,000 felt like a lot of money.

We started getting some traction, and I tried to hire somebody. I said, "We've got this company. It's called PRM Solutions." They said, "I have no interest in working at a company called PRM Solutions. That sounds lame."

We needed a new name. The ultimate idea was an AI system that could optimize all the profit levers in distribution. Sales, inventory, pricing, and rebates are inherently connected, but it's too hard for humans to coordinate all of them. Could AI connect those things?

That would be like a proton: a positively charged thing in the center of the distribution business. It turned out people wanted to work at a company called Proton AI.

In 2018 or 2019, no one knew what a .ai domain was. My uncle asked, ".ai? What is that? Don't you want a .com?" When we said, "It's proton.ai," people would ask, "Like proton.ai.com?" It ended up working out. At the time, I was just naive enough to think, "Maybe it'll work."

The pivots that brought Proton back to its first idea

Dasha: Proton wasn't the company it is today when it started. Walk me through the different shapes it has taken.

Benj: The first idea was to use a distributor's transactional and customer data to tell a salesperson who to call and what to sell.

Sales is a distributor's largest expense outside inventory. When you spend time with reps, you'll sometimes find that they're hanging out with their best friends, talking about the game, and not really selling. Could we help salespeople spend time on the highest-value work so they could make more money and the business could make more money too?

We proved that out in my family business with an inside sales team, then with other distributors. We also learned that our best customers were using the system like a CRM. Traditional CRMs didn't work very well for distributors. Distributors have lots of transactions and lots of products. Reps need to create quotes and orders. Traditional CRMs weren't built for that workflow.

We pivoted to an AI-native CRM focused on distribution, before "AI native" was a term. The business started to grow much faster.

The next iteration came with agentic AI. Proton began moving from a CRM toward an industry cloud that could connect these systems on one platform. Today, Proton has a product information management system, an agentic platform, and Order & Quote Entry Automation. We're building the software distributors need outside the core ERP.

Dasha: You make the levers of distribution sound stupid simple. How do they connect?

Benj: If you're a distributor, there are only a few ways to make more money. You can sell more stuff. You can raise your prices. You can buy better by purchasing the right inventory or getting better rebates. Or you can reduce OpEx by cutting spending on work you don't need.

All of those levers are related. If your sales team uses Proton, the system learns what's likely to sell and where reps should spend their time. That same AI engine should be good at forecasting what inventory to purchase. You buy more of the right stuff, sell more of the right stuff, and carry less dead stock.

Sometimes you get it wrong. You buy too much, or the market's willingness to pay changes. Inventory and sales data can feed a pricing engine that decides what to charge while balancing carrying costs and profit.

Rebates connect to all of it. You may change your pricing or purchasing based on a rebate. You may push a different product because it has a better margin or rebate.

Jeff Baker, who ran Platt Electric before it sold to Rexel, taught me this. Platt was widely known as one of the most profitable distributors. Jeff takes a complicated business and explains it simply: "If you want to make more money, these are the things you've got to do."

Software can now help distributors connect those decisions.

Sponsor: Most sales reps spend 72% of their week on admin tasks instead of actually selling. Proton changes that. We're a CRM built for distributors that guides your reps on who to call and what to sell. Our AI assistant, Pronto, even gives instant answers to questions like, "What do I need to know about this account?" or "Which product should I pitch them next?" Swish nearly tripled its growth rate by using Proton.

You could, too. Start by visiting proton.ai to learn more. Now back to the show.

The model release that changed how Proton builds

Dasha: There was an inflection point with AI that people outside our bubble might not have noticed. Talk about that moment and what has changed for us since.

Benj: Around Thanksgiving 2025, Opus 4.5 came out. It was the first large language model that was actually amazing at coding.

I remember coding with Sonnet, older versions of Opus, or ChatGPT. You had to copy and paste things. You had to look closely at the code. It was a lot of work. When Opus 4.5 came out, it was game over. It was an order of magnitude better. It crossed a threshold where you could automate real work.

I built many of our internal sales tools with it, including Proton IQ.

Dasha: Describe Proton IQ in a sentence.

Benj: Proton IQ helps us understand what's happening with our customers and in sales. It integrates Gong calls, CRM, email, and our knowledge base, then lets agents analyze all that data.

You can ask, "What do customers complain about with our mobile app?" The system can mine that data and pipe it into another agent that writes the code to fix the problem. Late in 2025, this was the first time you could get something like that fully into production.

It became clear that the models were going to automate a lot of coding work end to end. We changed our entire software development process. We also changed our product so customers could automate work more completely from beginning to end.

Why drafting an email misses the point

Dasha: A lot of people say, "I get AI. I use it to draft my emails. But I don't see the business transformation." Why is that?

Benj: Drafting an email misses the point. The idea behind Proton's agentic platform is that agents can do work a human does behind a computer. You can take a workflow from beginning to end.

The only way to understand that is to experience it. Our forward-deployed engineering and customer success teams work with distributors to deploy agents and show the magic for real.

Our own experience was the same. When we first used coding agents, we kept our software development process exactly the same and used AI to write code faster. We got a 20%, 30%, or 40% increase in output. That's not that interesting.

To get the full benefit, we had to overhaul the entire process, starting with how we listen to the customer. The product I built when Opus 4.5 came out is still the foundation for how we mine customer data to decide what to build. From there, agents carry the work through deployment, with a human managing and reviewing the process.

You don't get that result by adding AI to one step. You have to rethink the work for a world where agents can complete tasks end to end.

Why seasoned reps may adopt agents fastest

Dasha: There's a belief that older, less tech-savvy employees can't use or benefit from AI. What are you seeing?

Benj: I think that's totally wrong. Over the last 10 or 20 years, software packages have become more bloated. Companies keep adding features until the software is complicated to use. That's how you end up with people frustrated by Salesforce. You can do anything in Salesforce, and it's hard to do anything in Salesforce.

Agentic software changes the interface. In Proton, the agent can use the software to do what a human could do. Instead of learning where to click, a person can ask the agent, "Can you take care of these five things for me?"

You can add more capability while making the experience simpler. Some of the biggest users of our agentic products are seasoned reps because they already know what they want. They're not asking, "What should I ask?" They're saying, "I'm about to visit this customer. Here's the shit I need to know." The system gets it for them. They don't need to click around or open 10 different things.

Where should a distributor start?

Dasha: A lot of distributors feel overwhelmed. Their product data is bad, their sales team is reactive, and they rely on gut feel for inventory. Do they try to fix everything or roll things out in phases?

Benj: Do it in phases and do it iteratively. One of our values is "think big, start small." Think about all the ways you could use AI in the business, then start with one small use case. Prove the value. Use that value to pay for the next thing, and keep going.

Then it becomes a prioritization question. Is your product data really so bad that you can't tell the sales team who to call and what to sell? It's probably not so bad that you can't start there.

For agentic use cases, we've helped customers analyze their P&Ls to find where they're spending a lot on OpEx that could be automated. They can move people to higher-value work that matters more to the customer.

Start with a small use case, prove the value, and move on from there.

What distribution looks like 10 years from now

Dasha: What will still be the same in distribution in 10 years, and what will be different?

Benj: Distribution will fundamentally be the same. Distributors will still play a critical role between suppliers and customers.

What will change is the work people do. Ask what percentage of the dollars you spend in OpEx actually drives value for the customer.

A salesperson represents a lot of OpEx. Some of what a salesperson does is really valuable. Some of it isn't. Entering data into systems isn't valuable to the customer. Calling random vendors to track something down takes time. The answer matters to the customer; the time spent finding it doesn't.

I think we'll spend the same dollars, but we'll spend them on activities that make the customer experience way, way, way better. We'll look back and think, "I can't believe we used to spend money on tasks that didn't matter for the end customer."

The advice Benj only half believes

Dasha: As a founder, you get advice from a million different angles. What's an early piece of advice that turned out to be wrong?

Benj: "Work on the business, not in the business." I think that's great advice, but it's incomplete.

My experience has been that I learn the most by being super, super, super close to the problem and the customer. If I'm only working on the business, that doesn't capture where I learn best: visiting the customer, talking to the person closest to the work, and trying to understand the problem.

The advice is kind of right, but not all the way right.

What Benj is still curious about

Dasha: We end the podcast with a question about curiosity. What are you still curious about and haven't quite figured out?

Benj: I wasn't sure how deep to go on this question.

Dasha: Go deep.

Benj: I'm curious about how to feel satisfied in your life. So much of our culture, especially in the United States, says you have to make more money and that will make you more satisfied. My experience so far is that isn't really true.

I'm still trying to understand what gives me that deep sense of satisfaction. I suspect I'll be curious about that for my entire life.

Dasha: If you're listening and you have advice for Benj, please leave it in the comments. That's the pod.

Benj: That's it for this episode of In the Mind of a Distributor. If you liked this conversation, leave a review. It helps more folks in distribution find the show.

If you're thinking about a CRM for your business, head to proton.ai to see how Proton can help you grow faster. Or, if you want to connect with me, add me, Benj Cohen, on LinkedIn. Thanks for listening, and I'll catch you next time.