Aakar Khanna spent years watching his family's quick-service restaurants wrestle with the same problem: a tangled mess of back-office software that nobody actually used. Inventory counts done by hand. Purchase orders scrawled on paper. Schedules built in Excel and immediately obsolete.
So when Khanna and his co-founder Arjun Chaliha launched Marble in summer 2024, they pitched something ambitious. An AI operating system that would consolidate the whole sprawling back-office apparatus into one tool. The New York-based startup, backed by Y Combinator, says its system cuts food waste by 40% and frees up more than 30 hours weekly for managers running multiple restaurant locations across the U.S. and Canada—though these figures come from company reports rather than independent audits.
Bold promises, certainly. But the company is entering a market that desperately needs help. The National Restaurant Association estimates typical profit margins hover around 5%, with food and labor costs devouring two-thirds of revenue. Persistent inflation hasn't made things easier. Khanna describes the current software landscape as a "graveyard of unused logins," where operators toggle between separate platforms for counting inventory, placing orders, scheduling staff, and reconciling invoices.
Marble's pitch centers on computer vision and voice AI. Managers point their phones at storage shelves and speak item names aloud. The system verifies counts using object detection technology and then, according to the company, predicts demand at the item level with 97% accuracy. Machine learning models generate purchase orders, prep sheets, and schedules, though human approval is required before anything executes. Marble says the process runs 90% faster than manual counting, with 99% accuracy.
The platform hooks into major POS systems like Toast, Lightspeed, Square, Clover, PAR, and Revel. It connects with accounting software QuickBooks and Restaurant365, plus payroll providers Gusto and ADP. Marble's website suggests restaurants can go live within a week.
Khanna's background includes a stint covering restaurants and food and beverage as an investment banking analyst at Goldman Sachs. He later led strategy and operations at Kouper, a healthcare AI company, and worked in early business operations at AtoB, itself a Y Combinator alum from the summer cohort in 2020. Chaliha, who holds a CFA charter, previously led AI and machine learning initiatives for electronic trading at Bloomberg. The two met as freshmen at the University of Michigan's Ross School of Business and announced their Y Combinator acceptance around June, according to Khanna's LinkedIn.
Y Combinator's standard investment structure provides $500,000 in two tranches: $125,000 for 7% equity and $375,000 on an uncapped most-favored-nation SAFE, according to RuntimeWire.
Six customers appear on Marble's Y Combinator profile: A&W Restaurants, The Kati Roll Company, Sophie's Cuban, Kokomo NYC, HolyShred, and Dead Letter No. 9. Per Diem, a restaurant industry publication, profiled an A&W franchisee using Marble and reported inventory counting ran 90% faster—figures that align with company and customer claims. The piece noted demand forecasting accuracy near 97%, waste down nearly 40%, and stockouts "almost eliminated."
Marble declined to disclose pricing. The company does run a referral program offering $500 or an Oura Ring 5 when a referred restaurant goes live, a detail that suggests the startup is still in aggressive customer acquisition mode.
The competitive landscape is cluttered. MarketMan sells inventory and purchasing tools starting at $199 monthly. MarginEdge handles invoice processing and cost reporting. Crunchtime serves large restaurant groups with scheduling and kitchen management. StockCount recently added voice inventory and an AI assistant, with plans starting at $19 monthly.
Larger vendors have moved quickly in recent months. PAR Technology announced PAR Intelligence as an "agentic OS" for multi-unit operators in April. Oracle and NetSuite launched an AI-powered back-office suite in March. Nory expanded its "AI-native operating system" into the U.S. around the same time. Enterprise supply-chain platform Sightline OS emerged from stealth in May, and Fourth released a "Restaurant Operations Suite" in April of last year.
What stands out is how many players are racing to solve the same problem. The restaurant back office has long been fragmented, and AI offers a plausible path to consolidation. Whether any single vendor can actually unite the stack remains to be seen.
Marble's website features an anonymous testimonial from a QSR owner: "Marble is the first system we've ever used that demonstrably moved our profit number." The company previously operated under the name Truffle before rebranding, according to a banner on its homepage and third-party Y Combinator batch trackers. No additional funding beyond the Y Combinator deal has been announced.
For now, Marble's two-person team is betting that restaurant operators will embrace an all-in-one solution if it actually works. Given the industry's margins, they might not have much choice.
