Trent Duerr used to spend fifteen minutes crafting a quote at Milltown Paper, his distribution business in Neenah, Wisconsin. These days, he clicks once.
That shift—mundane as it sounds—is the wager at the heart of Arzana, a five-person startup that emerged from Y Combinator with a blunt diagnosis of American manufacturing: the problem isn't always on the factory floor. It's in the office, where humans still toggle between email, ERP systems, and spreadsheets, sometimes for hours, just to tell a customer how much something costs.
Arzana's pitch is straightforward, maybe deceptively so. The company sells what it calls an Office Execution System—software that deploys AI agents to handle the administrative work industrial firms do before anything actually gets made. Order entry. Quote generation. Customer service inquiries. The stuff that clogs inboxes and stretches response times from minutes to days.
Co-founders William Alexander and Marshall Kools, both Stanford graduates, claim their system processes requests ten times faster than manual workflows and cuts errors by 70 percent. Bold numbers, and in manufacturing software, bold numbers tend to require proof at scale. For now, Arzana says it's automating office work for more than twenty plants, though the company has been characteristically quiet about naming customers beyond a handful of case studies.
The Execution Play
The term "Office Execution System" is deliberate positioning—a riff on Manufacturing Execution Systems, the software layer that orchestrates machines on the shop floor. Alexander and Kools argue that front-office workflows need the same kind of orchestration. Quoting, order intake, vendor outreach: these aren't one-off tasks but interconnected processes that break down under volume.
So Arzana built what it describes as an "arsenal" of agents. One pulls request-for-quote emails, extracts specs, matches them against inventory or capabilities, then generates a quote. Another converts purchase orders from email into ERP entries. A third monitors customer service inboxes, responds to order status requests, pulls tracking details from backend systems. It's automation, but packaged as a bundle rather than point solutions.
The platform integrates with the usual suspects: SAP, Oracle NetSuite, Epicor, Microsoft Dynamics, plus CRM systems like Salesforce and HubSpot. According to the company's FAQ, most integrations finish in under 48 hours, though implementation timelines elsewhere on the site stretch to 120 days. That gap—between the promise of speed and the reality of enterprise software deployment—will matter as Arzana scales.
The architecture, outlined in a Y Combinator job posting for founding engineers, isn't exotic. Email ingestion. Document extraction via vision-based large language models. Record matching from ERP and CRM systems. Agent orchestration. Engineers build discrete agents that chain together into workflows, and customers can customize those chains with what Arzana calls "prompt-based" configuration. Whether that's genuinely flexible or just marketing speak depends on how gnarly a manufacturer's workflows get.
Milltown and the One-Click Quote
Duerr's experience at Milltown Paper shows up as Arzana's published case study. The company switched from legacy software and implemented custom RFQ matching across more than 211,000 inventory rows. Quotes that took fifteen minutes now generate with a click, Duerr said in a testimonial. It's the kind of efficiency gain that sounds incremental until you multiply it across hundreds of quotes a week.
Arzana's website displays logos from other industrial firms—Tecton, Iowa Mold & Engineering, Schneider—though independent confirmation of these relationships wasn't available. In one LinkedIn post, the company noted that a mold quoting system for Iowa Mold & Engineering reduced quote time from one to three hours down to one to three minutes. If accurate, that's a 60x to 180x improvement, the sort of compression that either transforms a business or reveals how broken the old process was.
The company advertises 99 percent accuracy or higher and says the system can go live in 30 days. Pricing starts somewhere between $2,000 and $4,500 per month, according to the FAQ—a range that suggests flexibility or, perhaps, that Arzana is still figuring out what customers will pay.
A Crowded Space, With Room to Maneuver
Arzana isn't alone in this corner of manufacturing software. Paperless Parts has focused on quoting for job shops, with AI-assisted estimating and ERP integrations. Faction offers modules for quoting, order entry, pricing, procurement. HublerX provides an RFQ-to-order operations platform with intelligent pricing.
What Arzana seems to be betting on is breadth. Rather than forcing manufacturers to stitch together multiple vendors—one for quoting, another for order entry, a third for customer service—the company bundles it all into a single system. That's either a strength or a liability. Breadth can mean comprehensive; it can also mean unfocused. A March analysis from Aimyflow, an industry review site, suggested buyers validate workflow readiness across Arzana's wide surface area before committing. Fair advice, if a bit obvious.
The broader trend is clear enough. AI-driven execution layers are proliferating across manufacturing. L2L launched Execution AI for shop-floor intelligence. Invisible AI unveiled its Vision Execution System at Hannover Messe. These products target production and quality monitoring rather than office work, but the convergence on "execution" framing—systems that take action, not just analyze—signals something shifting. Dashboards are table stakes now. The software that wins will be the software that does.
Team and Trajectory
Arzana was founded in 2025 and went through Y Combinator's Spring 2026 batch. Aggregator databases like CB Insights and Dealroom list funding amounts, though the company hasn't issued a press release confirming specifics. The team of five operates from San Francisco, with additional offices in Okoboji, Iowa, and Appleton, Wisconsin. Alexander is CEO, Kools is COO. The company has posted openings for founding engineers, listing compensation in the $70,000 to $105,000 range plus equity. (Kools was featured in a February New York Times article titled "These A.I. Dreamers Don't Fit the Stereotype," which is either good PR or a sign of something interesting in his background.)
The challenge for Arzana—and it's not a small one—is proving that a single platform can handle the variability of industrial workflows. Job shops that quote custom tooling operate differently than distributors moving catalog inventory, and both differ from paper converters or trucking brokers. Arzana lists all these verticals as targets. That's ambitious. Maybe too ambitious.
Still, the product is live. Customers are using it. The company is recruiting engineers to expand the agent library. Whether the Office Execution System becomes a category or gets absorbed as a feature set by larger ERP vendors will depend on how well Arzana's agents perform when order volumes spike and edge cases multiply—the moments when automation either proves itself or breaks spectacularly.
For now, manufacturers like Duerr are clicking once instead of spending fifteen minutes. If that holds at scale, Arzana might be onto something. If it doesn't, well, there's always another startup waiting to automate the front office. There always is.
