Meet proton.ai, Your New A.I. Sales Assistant
I recently had the chance to interview Benj Cohen, a former teammate of mine and now the Founder & CEO of proton.ai, a Cambridge, MA-based AI company focused on helping sales teams sell intelligently. This conversation has been edited for length and content.
Benj, thanks for taking the time. So to start off with, can you give me a bit of context on what Proton AI is doing and who it’s working with?
At Proton we build software for wholesale distributors. Our customers sell medical supplies, industrial supplies, and all kinds of different things that help the world run a huge range of businesses. We use AI to help those companies figure out what their customers need to buy, and then we help their reps market and sell those products. It’s kind of like Amazon’s suggestive selling online, but instead of just doing it for a website, we do it for all of a distributor’s go to market channels.
Good stuff. Can you walk me through how Proton came to be?
I grew up in wholesale distribution. My family has been in the dental distribution business for generations, so I grew up in that world and loosely planned on staying in the family business. But when I went to college and started studying applied math and data science, I got really interested in how AI could solve problems in my family business. So, the summer after my junior year of college, I got some friends together and we lived at my house for the summer and built the first version of our product for my family business. And it worked! After that we were like, oh man, we should do this for other distributors and that’s how we got started.
Since Proton was founded, how has your concept changed? Are you still solving the same problem?
I think the core challenge that we’re solving for is really the same; we help distributors grow their businesses faster by using AI to turn passive sales channels into active selling avenues. Our product has changed in a lot of smaller ways, as we’ve tweaked the approach for different channels. But, the core idea is still the same.
Can you flesh out for readers what exact service Proton is providing to distributors?
Proton helps create something that we call the distributor AI Flywheel. We start by pulling in all the data we can: Information from ERP’s, product data, pricing data, e-commerce data, and CRM’s if our customers have one. After that, we pass all of that data into Proton’s engine, which uses AI to predict who you should call and what you should sell. Not only do we figure out what people are likely to purchase, but we also figure out what sequence of events needs to happen for that purchase to happen.
That output is different in each channel. For example, in e-commerce and marketing channels we provide product recommendations on websites and in targeted emails that help personalize a customer’s digital journey. But for sales channels, the goal is to make sure that reps are pitching the right products to the right customers at the right time. For order taking reps, this looks different too. In customer service, for example, we provide suggestions on potential upsell and cross-sell opportunities for every customer, in real-time.
The best part about our system is that we are able to coordinate all of this activity across channels. For instance, if a potential customer views something on a client’s website, that’s an opportunity for a sales rep to follow up with them and secure that sale. At the end of the day, we want to make sure that every customer gets relevant product offerings and good service every time they interact with a distributor, no matter the channel.
In terms of integrating with third parties, how connected is Proton now, and how connected do you think Proton will be in the future?
Some of our larger customers use CRM’s, and we can integrate with those platforms directly. We also have multiple APIs that customers can use on their websites and in their marketing applications. Other customers don’t use a standard CRM can use our web application as a CRM. In the long term, we envision our platform being the single place sales reps go to plan their day.
Let’s talk a bit about what it’s like to work with your clients. So let’s say you were rolling this tool out at a new client, what would that process be like from start to finish?
We put together a six-week project plan for each customer. We start working with the customer right away, as there are a couple of days of data integration to do upfront. Once we’ve pulled in and cleaned the data, we basically set up the system on our side. While our programs are learning, we work with the customer's reps to do quality assurance and build an onboarding plan. In total, that takes about six weeks, which is fairly quick as far as enterprise software goes.
I’m assuming that most of your customers probably haven’t used a tool like Proton’s before, what are some of the challenges when you’re implementing your tool and delivering these recommendations to your clients?
The big challenge with delivering any system is making sure that people buy-in and actually use it. Of course, the product has to actually add value, but after that, it’s a change management exercise. We work hard to align all the layers of management around the initiative. It’s great if a CEO is bought in, but it’s not going to work unless the middle managers and reps are also aligned on the why behind the strategy. So it really just helps for reps to know why our product is beneficial for them. And that is because it simplifies rep workflow and helps them increase sales to earn more on commission.
What are some of the largest challenges for Proton when it’s collecting data from clients?
We don’t have a ton of challenges on that front. We work very hard to help people find the data that we need and get it to us efficiently. Plus, we’ve built models that accept complicated or unstructured data. And, we have an implementation team that converts data into a workable format. We’ve worked hard to design a system that is very flexible with data, and the payoff has been that complicated data has never been a showstopper for us.
So some companies do a great job maintaining their data, others not so much. Why do you think some companies fail to prioritize data hygiene?
In general, I don’t think distributors give themselves enough credit for the cleanliness of their data. Companies who are setting up e-commerce platforms from scratch tend to make data hygiene a priority. For distributors, the data hygiene challenge is mostly related to the high number of products and customers they serve.
Are you using third party data sets beyond what your clients are providing to make recommendations?
Yes, we do. We use demographic and firmographic data, which is really valuable when we’re trying to solve the cold start problem (when a customer is purchasing for the first time). If you’ve got a customer who’s never bought anything before, 3rd party data can help make a recommendation for them.
So post rollout, how does Proton track the value it’s delivering to clients?
We use an attribution system that links activity in Proton to sales. If a rep talks to a customer about a product that we recommended, and then that customer buys the product, that counts as a point for Proton. We’re really careful to track that because we want to make sure that our clients can see the immense value that we deliver. We also give senior leadership access to the software so they can log in themselves and track performance.
What have been some of the challenges that Proton’s faced when working in this space?
I think the biggest challenge with any new tool or piece software is that the tool itself is not the ultimate answer. The ultimate answer is changing people’s behavior in order to make a difference. So I think the biggest challenge today is getting people to make that behavior change so that they can actually use the tool to create value for their firm and their customers.
How do you get buy in to make sure that behavior change happens?
We focus on adding value for reps. First, we help reps make more money. Second, we save them time while offering customers a better experience. We feel that there’s strong value there, and it’s something we reiterate over and over. We also work really hard to set the system up properly for everyone on the team and coach folks to use proton to its full potential.
So given your experience launching a company in undergrad. What advice do you have for other startups/students/researchers that are also trying to solve problems with data?
My experience has been defined by not knowing. I’d never built a product or a team, sold an enterprise customer, etc. The three things that have helped me navigate this unknown territory have been (1) building a super sharp (and fun) team, (2) seeking lots of advice from mentors, and (3) trying things out (test and learn).
What excites you the most about Proton’s future?
Well, first, I’m really excited about getting to work in the market that we’re in, I like working with distributors. We believe this space is underserved, so we’re excited about offering services and creating opportunities that didn’t exist before. I’m also excited because our product adds clear and measurable value to our customers quickly. It’s always fun to work on something that you know is making a difference. Lastly, we’ve got great people, and that makes it fun for all of us to work together.
