Steph Curry’s investment venture, Penny Jar Capital, is leading the $3.8 million seed round in Burnt — a Y Combinator company using AI agents to codify and help manage this chaotic backend of food supply chain operations.
The bet falls into an area where orders continue to come in through emails, texts and faxes, and where enterprise software rollouts remain all too expensive, sluggish and underused.
Burnt’s pitch is simple: let AI tackle the repetitive workflows that swamp distributors and importers without requiring teams to rip out systems that are already running their businesses.
The company says its software already handles $10 million in monthly orders and generates six figures in revenue, with one U.K. food conglomerate currently rolling out the platform.
Why Food Supply Chains Need To Be Automated
The food economy is vast and unforgiving. According to industry groups like FMI and the National Restaurant Association, U.S. grocery and foodservice sales total well over a trillion dollars a year, with scant room for clerical errors, late invoices or interrupted traceability.
But many distributors are still juggling phone calls, PDFs and spreadsheets in order to take orders and keep their inventory flowing. It’s that fracture that is driving cost and waste. The U.S. Department of Agriculture estimates that approximately a third of the national food supply is wasted; even marginal improvements in accuracy and speed could lead to billions saved across procurement, fulfillment and compliance.
ERP systems are everywhere, but not for long. Distributors often describe them as essential infrastructure that is too expensive and dangerous to replace. Burnt’s argument is that for many companies, AI agents can sit on top of those tools, orchestrating data entry and exception handling throughout email, messaging apps and EDI without undertaking a complete replatforming.
The Making of Burnt And Its AI Agent Ozai
Burnt’s initial agent is named Ozai, and it emphasizes order intake and entry — a seemingly simple but actually time-consuming task that can gobble up hours of staff members’ time on a daily basis.
Think of it as an AI clerk that reads unstructured requests (which come with the industry territory), matches those requests to SKUs and pricing, spots discrepancies, and then dumps clean data into whatever ERP a given distributor is already using.
The company says its system, called Ozai, can handle up to 80% of the process steps that workers deal with, escalating edge cases to people after learning from each resolution. That blend of independence and oversight is especially important in food, where the wrong substitutions or forgotten allergy warnings pose real threats.
Founder and CEO Joseph Jacob has deep domain roots: His family has worked various points in the seafood value chain for generations, and he helped build software for restaurant suppliers at Rekki previously. Rounding out the team are co-founders Rhea Karimpanal, a product leader who grew up in restaurants, and CTO Chandru Shanmugasundaram, a hospitality systems industry vet; together they make for a team that speaks the language of restaurant life.
Why Penny Jar Capital Is Leaning Into Food Distribution
Penny Jar Capital has focused on founders going after underserved categories where tech adoption is low. Food distribution hits all the marks: big market, established processes, and resistant buyers who remain skeptical after years of expensive software projects that never materialized.
The round is led by Scribble Ventures, with participation from Formation VC and angels like longtime enterprise investor Dan Scheinman. Their bet is that AI agents can drive hard ROI — fewer keystrokes, faster order-to-cash, and lower error rates — without the multi-year change management that derails so many ERP replacements.
Regulatory and competitive context for food AI adoption
Compliance pressure is rising. The FDA’s Food Safety Modernization Act traceability requirements are driving distributors toward digitized records and standardized data capture. Organizations like GS1 US have been advocating for more far-reaching use of data standards in order to enable end-to-end traceability, recalls and audits. AI that organizes orders and metadata at the time of intake can help make compliance a little less like a fire drill.
The competitive group is out there but splintered. Afresh applies AI to grocery forecasting, Crisp focuses on demand signals for CPGs, and visibility platforms including project44 and FourKites track shipments in motion. Burnt is carving out the unglamorous but critical back-office lane inside distributors — the work no one sees but on which everyone relies.
What to watch next as Burnt expands its AI platform
Burnt’s initial test is breadth: expanding beyond order entry into adjacent workflows like credit memos, chargebacks, inventory reconciliation and claims. Each has different failure modes and data quirks, and the future will depend on guardrails that prevent models from veering off-track in high-stakes environments.
The upside is compelling. A small reduction in manual entry and exception churn can also recover thousands of staff labor-hours per distributor per year, help you increase fill rates, and expose real-time margin insights. Should Burnt be able to turn those gains into playbooks it can repeat across seafood, specialty goods and broadline categories, Penny Jar’s bet may age well.
For an industry that operates on early mornings, thin margins and deep relationships, the technology that will win won’t be flashy. It will be the silent operator that sweeps inboxes, tidies up data, spares humans to sell and serve. Part of what Burnt is trying to do with automation — one order at a time.
