About Us

Proton is building the AI operating system for wholesale distribution, embedded in the workflows that move nearly every physical product on the planet. Distribution is a $9 trillion industry running on software that has been stuck for decades. We unify CRM, PIM, eCommerce AI, and Order & Quote Entry AI into one platform with one data layer and one AI brain, so reps spend their time on customer relationships instead of data entry.

We have been building AI into distribution since 2018, long before it became an industry buzzword. That head start matters, and we've only deepened the investment: in 2025 we shipped the first AI assistant embedded in a distribution CRM and launched a complete agentic product in our PIM offering.

We hire people with high agency and high urgency. Everyone here is a builder who owns problems end to end and ships AI-native software faster than anyone else in the category. The median Proton customer reports 3x profit return per dollar spent and an extra day of sales per rep per month. Hundreds of distributors run on Proton, from family-owned shops to publicly-traded enterprises. We're backed by Felicis Ventures and Battery Ventures.

The Role

We're hiring a Product Manager to own product domains end to end. Reporting to the VP of Product, you'll lead a multi-functional squad at the intersection of AI capability and real distributor workflows, turning our positioning advantage into compounding product differentiation.

End to end means further than it usually does. You inherit a problem space rather than a backlog: work out where the real value sits across hundreds of distributors, ship in tight cycles with design partners keeping you honest, prove the thing works in production, then drive adoption across the base and make it something sales can sell without you. There is no handoff point where the domain stops being yours.

Location: Cambridge, MA. Hybrid, 3 days a week in our Harvard Square office.

What You'll Do

Own your domain from fuzzy to growth. Take an ambiguous problem space through discovery, definition, build, launch, adoption, and revenue. You are responsible for the outcome.
Get onsite, constantly. Distributors work in warehouses and on sales floors. Onsite visits, shadowing sessions, and discovery calls are how you learn the domain, and the best calls come from standing next to a rep. Our product people know their users by name.
Run genuinely iterative development with customers in the loop. Think big, ship small, put working software in front of real users early, and let what you learn reset the next cycle. Design the feedback loop deliberately rather than waiting for feedback to arrive.
Build hands-on with engineering, design, and data science to ship features and agentic workflows, and to make sure the surrounding UX lets a user act on what AI surfaces.
Own adoption, not just launch. Run alphas and betas, stay close through the first rollouts, then push adoption across the installed base and watch where it stalls and why.
Enable GTM. Give sales and CS the positioning, proof, and training they need. Join customer and prospect calls, show the product yourself when it helps, and make sure what you built is sellable and supportable without you in the room.

What You'll Need to Succeed

Product and startup experience. At least 2 years managing features end-to-end through the SDLC. Early-stage startup experience (Seed through B or C) is a plus.
Real agile experience, and the judgment to adapt it. You've worked inside traditional scrum, know what those rituals were for, and can run a tight iterative loop with customers rather than a release calendar. You've shipped something incomplete on purpose because it was the fastest way to learn.
Technical and AI collaboration. You've built alongside engineers and data scientists to deploy AI-centric features at scale, and you're fluent enough to hold your own with senior engineers on data models, APIs, and service boundaries.
Daily, hands-on use of AI in your own practice. Claude Code, Cursor, or equivalent, used to ship real work rather than generate artifacts. You can say concretely how you structure context and how you check what comes back.