Aakar Khanna spent his childhood counting inventory in his family's restaurants. He went to Goldman Sachs, then to two startups, and finally back to the problem he knew best. Marble is what he built.
Every restaurant runs on two numbers that nobody in the dining room ever sees. One is inventory - the food sitting in the walk-in, the case of tomatoes that spoils before it sells, the order placed too late or too large. The other is labor - who is scheduled, for how long, and whether the shift matches the crowd. Get those two numbers wrong for long enough and the restaurant closes, no matter how good the food is. Aakar Khanna learned this before he could legally drive.
He grew up inside his family's multi-unit restaurant business, the kind of operation where the margins are thin and the math is relentless. That is not a metaphor he reaches for in a pitch deck. It is the actual texture of his childhood - counting stock, watching labor costs, absorbing the quiet stress of a business where a bad week of forecasting shows up directly in the bank account.
Most founders study a problem from the outside and then work to understand it. Khanna started on the inside and spent his career slowly circling back.
Khanna studied at the University of Michigan's Ross School of Business, adding a minor in Computer Science - a small choice that reads differently now that he runs a software company. At Michigan he played club soccer, joined the Zell Early-Stage Fund, worked with 180 Degrees Consulting, and spent time in the Alternative Investments Club. The pattern was already there: business, building, and a pull toward early-stage things.
After graduation he landed at Goldman Sachs as an investment banking analyst. His coverage area was almost poetic - Restaurants and Food & Beverage. He was analyzing, from a spreadsheet on a trading floor, the exact industry his family ran by hand. He saw the sector from the capital markets side: the multiples, the unit economics, the reasons some chains scaled and others stalled.
But finance is a place you observe a business, not fix it. Khanna moved to the operator's chair.
He became an early business operations hire at AtoB, a YC Summer 2020 company - the kind of role where you do a bit of everything because there is not yet anyone else to do it. Later he served as Head of Strategy & Operations at Kouper, an applied AI company. Between the two he collected the two things a first-time founder usually lacks: how to run the guts of a fast company, and a working intuition for what AI can actually do rather than what it is marketed to do.
By 2026, the threads had braided together. A childhood in restaurants. A finance education in the sector. Years of startup operations. A minor in computer science that suddenly mattered. Khanna teamed up with Arjun Chaliha - a former Bloomberg engineer, CFA charterholder, and fellow Michigan grad - and started building.
The company launched as Truffle and later took the name Marble. The mission stayed identical through the rename: an AI-native operating system for the restaurant back-of-house. Not the point-of-sale, not the delivery app, not the dining room - the unglamorous machinery behind the kitchen door where the money actually leaks.
The pitch is refreshingly narrow. Restaurants juggle a stack of disconnected point solutions, spreadsheets, and literal clipboards to manage inventory and labor. Marble replaces that mess with one system. Computer vision and voice AI handle inventory counts - the company claims those go about 90% faster than counting by hand. Machine learning forecasts demand, with a reported 97% accuracy. Autonomous agents then handle the follow-through: procurement, scheduling, invoice reconciliation, even menu engineering. The manager stops doing the counting and starts approving the decisions.
Figures reported by the company for early deployments.
Marble's early customer list mixes a legacy chain with a slate of New York favorites: A&W Restaurants, The Kati Roll Company, Sophie's Cuban, Kokomo NYC, HolyShred, and Dead Letter No. 9. It is a telling spread - a national brand that needs to standardize across units, alongside independent operators for whom a single wasted case of produce is a real dent. Both feel the same two pressures. Both are the whole thesis.
Relative time to complete an inventory count, per Marble's ~90% faster claim.
There is a detail in Khanna's history that explains more than any funding round. In 2016, while still a student, he founded Play it Forward Academy - a pro-bono soccer training academy in the Greater Toronto Area. He ran it for seven years and led a team of 16 volunteers. It made no money and was never meant to. What it shows is a person who builds durable, team-run systems for their own sake, and keeps them alive long after the initial excitement fades. Running an autonomous operating system for restaurants is, in its own way, the same instinct pointed at a harder problem.
Everyone wants to disrupt the dining room. The money leaks in the back.
The bet Khanna is making is simple to state and hard to build: restaurants do not need another app bolted onto an old workflow. They need one system that counts, forecasts, orders, and schedules on its own, and gets out of the operator's way. The unglamorous back-of-house, not the shiny dining room, may turn out to be where AI matters most in food - because that is where the hours and the dollars quietly disappear.
Khanna has spent his whole life around that disappearance. Now he is trying to stop it, one walk-in freezer at a time.