Boston-based supply chain startup Atomic has raised a $12.5 million Series A round led by Klass Capital and Madrona Venture Group, bringing total funding to just over $15 million. The company, which counts DoorDash and HelloFresh among its customers, has quintupled annual recurring revenue since January and added longtime Tesla planning director Jeff Goodrich as CTO and third co-founder.
Atomic's software determines how much inventory a company should hold and where, simulating scenarios and recommending - or automatically executing - responses. The approach traces back to the 2018 Model 3 production ramp at Tesla, where the founders built an early version of the system after spreadsheets failed to keep pace with rapid planning changes.
From recommendations to autonomous decisions
CEO Michael Rossiter said the company has moved beyond pilot programs. "The company has transitioned from pilot customers into real customers, and by real customers, it's DoorDash-scale customers," said Jon McNeill, a former Tesla president and founder of DVx Ventures, where Atomic was incubated. McNeill also sits on Atomic's board.
DoorDash now runs roughly 90% of its purchasing across hundreds of sites through Atomic's platform, according to McNeill. The product has evolved from offering recommendations to making decisions autonomously. For food-focused customers, that translates directly to less waste and spoilage.
"If you think about running a supply chain, running an operating model, it's like an infinite search space for optimization that you're trying to figure out all the decisions you could make at any given time - and then it changes all the time too," Rossiter said. "AI can play the role of finding all of the best paths through that forest."
Fast deployment across industries
Atomic's ability to onboard customers quickly was a key factor in investor interest. McNeill said the board challenged the team to compress onboarding time, and co-founder and Chief Product Officer Neal Suidan led the effort. Suidan pushed the agentic AI to infer "decision rules" that customer staff followed even when those rules weren't documented anywhere.
"Then customers were saying, 'Okay, then you might as well make the decision and free my time up,'" McNeill said. He pointed to decision speed as a core principle from his Tesla years. "Decision speed compounds. Like, I make a decision today, I build on that decision tomorrow."
Rossiter said the software adapts across sectors. Current work spans consumer packaged goods, mobility, and manufacturing clients. "The cool thing about Atomic is it's a general model of how you think about supply chains and operating models, and so no matter what that system looks like, our AI can adapt to it and tailor fit it," he said.
Moving operations data out of spreadsheets
Rossiter framed Atomic's value as pulling supply chain planning out of manual spreadsheets and into software that can handle planning decisions at speed. He said CFOs have led similar shifts for finance functions, but operations data rarely gets the same treatment. "Finance data always gets the priority. Operating data doesn't always get that," he said.
For operations leaders weighing AI adoption in inventory and planning, the pattern here is clear: customers are not just getting dashboards or suggestions. They're handing over execution. That's a meaningful shift in trust and workflow design, and it's where AI for Operations Courses can help teams understand what autonomous decision-making means for their own processes. Production planners specifically may want to explore AI Production Forecasting Courses to build the skills needed to evaluate tools like this.
Why this matters for operations and construction professionals
Atomic's growth signals a broader shift: companies are moving from AI that advises to AI that acts. For operations, real estate, and construction professionals, the pressure to match decision speed will come from competitors who adopt similar tools. The takeaway is not to automate for its own sake, but to identify which recurring planning decisions - inventory levels, material orders, site allocations - currently depend on tribal knowledge or outdated spreadsheets. Those are the decisions an agentic system can learn and execute, freeing staff for exceptions and judgment calls that genuinely require human input.
Your membership also unlocks:
