There's a particular kind of tedium that haunts American manufacturing offices. An email arrives with a purchase order. Someone opens it, squints at the part numbers, manually rekeyes everything into an enterprise system that looks like it was designed during the Clinton administration. Mistakes get made. The process repeats itself, again and again, hundreds of times a month.
Arzana thinks it has a better way.
The San Francisco startup—backed by Y Combinator—has raised seed funding with participation from Zero Index VC, confirmed earlier this year, though the company has kept quiet about the total amount and other investors.
The pitch sounds almost prosaic until you consider the scale: Co-founders William Alexander and Marshall Kools, both Stanford alums, estimate that manual order processing drains 6 to 10 percent of revenue from the average manufacturer. That's not a rounding error. That's money simply evaporating into administrative friction.
A Layer Between Email and Enterprise Software
What Arzana calls an Office Execution System—OES, inevitably—is essentially a buffer zone of automation. The software sits between incoming communications and legacy ERP platforms, ingesting quote requests and purchase orders from email or PDF, cross-referencing parts catalogs, validating pricing against existing customer agreements, and flagging anomalies before pushing clean data into systems like Epicor, SAP, or NetSuite.
The company claims integrations with at least nine major ERP platforms, including JobBoss, Infor, and Microsoft Dynamics. According to Arzana's materials, the system delivers processing speeds ten times faster than manual entry and cuts errors by 70 percent. The company says implementations typically wrap in under four months.
These are company-supplied figures, of course. Independent verification remains elusive.
Still, early customer testimonials suggest genuine traction. Milltown Paper, an early adopter, highlighted one-click quote generation in a public endorsement. Founder posts on LinkedIn reference on-site deployments at Iowa Mold & Engineering and engagements across the printing and converting industries.
Job Shops and the Complexity Problem
Arzana isn't chasing mass production lines. Its sweet spot is messier: job shops, mold makers, make-to-order operations, and wholesale distributors—the kinds of businesses where every order is a little different, and standardization is more aspiration than reality.
It's a smart wedge, maybe. Custom quoting and variable order complexity make these segments particularly resistant to off-the-shelf automation tools. The company operates lean, with a team hovering around five or six people split between San Francisco, Okoboji, Iowa, and Appleton, Wisconsin. That geographic spread isn't accidental—it mirrors the Midwest manufacturing base Arzana is courting.
The timing looks deliberate too. Over the past year, manufacturing operations software has seen a flurry of new entrants. Proton.ai rolled out its own order and quote automation platform targeting distributors. Paperless Parts continues to build out CAD-driven quoting tools for precision manufacturers. Arzana's angle—positioning the OES as a customizable layer rather than an ERP replacement—is essentially integration over disruption. Play nice with legacy systems rather than attempt to rip them out.
Whether that's strategic discipline or tactical caution is hard to say.
The Staffing Math and What Comes Next
The company claims to have saved customers more than $1 million in avoided staffing costs—a figure that, again, originates from company materials and comes with all the usual caveats about self-reported ROI. But the math isn't hard to sketch out: if you're processing hundreds of orders a month and each one requires 20 minutes of manual data entry, the labor savings add up quickly.
The company has been hiring. A recent job posting for a Member of Technical Staff role in New York City offered $150,000 to $170,000 in salary plus equity between 0.25 and 0.50 percent—generous by startup standards, if not quite FAANG-level compensation. With the seed round closed, the focus appears to be scaling engineering capacity and expanding customer deployments.
Earlier this year, The New York Times profiled co-founder Marshall Kools as part of a cohort of AI founders who don't fit the stereotypical Silicon Valley mold—a reference, presumably, to the company's Midwest manufacturing connections and multi-city footprint. It's the kind of narrative framing that plays well in an industry often accused of coastal insularity.
The Institutional Knowledge Question
But here's the thing that keeps nagging: Can AI agents truly replicate the institutional knowledge embedded in a manufacturing office veteran who knows, instinctively, when a part number looks wrong or when a customer's quoted price feels off?
Arzana's early results suggest the technology can handle the rote work—data entry, invoice matching, routine customer service queries. The trickier question is whether it can absorb the edge cases, the judgment calls, the moments when experience trumps algorithms.
For now, at least, Arzana has capital, early customers willing to test the premise, and a market opportunity measured in percentage points of revenue across an entire sector. Whether that's enough to build a durable business—well, that's the bet.
And like most bets on automation, the answer won't arrive neatly packaged in a quarterly earnings report. It'll show up in factory offices, one eliminated spreadsheet at a time.
