The rituals of industrial distribution haven't changed much in thirty years. A customer emails a parts list. Someone pulls up a clunky ERP, hunts down SKUs across multiple tabs, matches pricing by hand, generates a quote. Repeat, dozens of times a day. The whole process moves at a pace that would feel archaic in almost any other industry—but this is how millions of parts get bought and sold.
Hexa, a tiny startup fresh out of Y Combinator, thinks AI agents can finally change that. And maybe they're right, though the company's early traction—three paying customers, $8,000 in monthly recurring revenue after roughly six weeks—is more suggestive than conclusive. Still, in an industry where software adoption typically moves at glacial speed, any early momentum is worth noting.
The San Francisco team, currently just three people, is building what it calls "The OS for Manufacturers and Distributors." That's ambitious framing for a company still in its infancy. But the problem they're tackling is real: industrial distributors are drowning in manual work that computers should have automated years ago.
What Hexa Actually Does
The platform divides into four modules, each targeting a different operational bottleneck. Sales agents generate quotes directly from emails, PDFs, or phone calls—pulling catalog data, calculating pricing, and pushing finalized quotes without anyone needing to crack open the ERP. Procurement agents flag material shortages, compare vendor pricing, send out requests for quotes, and create purchase orders autonomously. Finance agents reconcile invoices against purchase orders and receipts, track aging receivables, and handle collections outreach. Customer service agents field questions about order status, delivery windows, product specs, and returns using live data from the ERP, only escalating when they hit the limits of what software can handle.
It's integrated with the systems distributors actually use—Prophet 21, SAP, NetSuite, Infor, Epicor, Microsoft Dynamics, Salesforce, HubSpot, Zoho, Pipedrive. According to founder and CEO Ishaan Makkar, these aren't copilots that assist humans. They're agents designed to complete tasks independently.
The deployment model is straightforward: Hexa sends people on-site to shadow workflows, identify where automation can replace repetition, then deploy and tune the agents over time. It's a labor-intensive approach that might not scale easily, though at three customers, scale isn't the immediate problem.
A Crowded Race
Here's where things get interesting—or concerning, depending on your perspective. Hexa is far from alone.
PhaseZero launched eight AI agents for manufacturers and distributors on May 12, 2026, emphasizing Model Context Protocol integration for parts identification and quoting. Canals introduced its "Operating AI for distributors" in April, complete with live voice capabilities for real-time phone order entry. Prokeep launched AI-powered order automation at the NAW Innovators Summit on September 16, 2025, ingesting everything from text messages to Excel sheets to generate ERP-ready quotes.
Even within Y Combinator, Hexa faces near-clones. Avent, Comena, and Seals AI all launched recently with strikingly similar positioning: AI agents for industrial commerce, order processing, wholesale distribution. The language across these pitches is almost interchangeable, which raises an obvious question about how any of them differentiate beyond execution.
The established players aren't sitting idle either. Zilliant launched agentic AI with Model Context Protocol capabilities targeting early 2026 availability. Epicor has been embedding generative AI directly into Prophet 21, the widely deployed distributor ERP. Conexiom and Reshape Automation both rolled out AI-powered platforms last year.
There's clearly a market here. There are also a lot of hands reaching for it at once.
The Industry Might Actually Be Ready
Perhaps the timing is finally right. Distribution Strategy Group's March 2025 survey reported rising AI adoption among distributor teams. Industrial Distribution's June 2025 survey showed 29% of distributors using AI—though only about 10% reported measurably positive impact so far, a sobering reminder that hype and results don't always align.
The National Association of Wholesaler-Distributors launched its own AI assistant for the sector. Industry conferences like MDM SHIFT in May centered entire agendas around AI adoption and hybrid sales models. The conversation has moved from "should we?" to "how do we make this work?"
The Test Ahead
Makkar founded a B2B ed-tech startup before this that reportedly scaled to over 70 partners and 5,000 users, according to his Y Combinator profile. He studied physics at Imperial College London. That background suggests pattern recognition and systems thinking, both useful when trying to impose order on industrial chaos. Whether it translates to building a durable software business in a relationship-driven, change-resistant industry is another matter.
For Hexa, the challenge isn't demonstrating that AI can automate quotes or purchase orders—competitors are already proving that much. The challenge is proving it can do so reliably enough, cheaply enough, and integrate tightly enough with decades-old systems to earn trust in an industry where a single botched quote can end a customer relationship.
Three customers in six weeks suggests the pitch resonates. Whether the agents perform consistently enough to retain those customers—and whether the economics work at scale—is what comes next. The industry's ready for better tools. Whether Hexa becomes one of them, or just another name in a very crowded field, depends entirely on execution from here.
