Somewhere in a metal finishing shop right now, someone is probably squinting at a smudged purchase order that arrived by fax. Or worse, handwritten.
It's a workflow straight out of 1995, and Sam Goldman and Taira Fujioka think they can fix it. The two engineers, working out of San Francisco, emerged from Y Combinator's Spring 2026 batch with Smartbase—a piece of software that does one thing, but promises to do it exceptionally well: turn illegible purchase orders into clean data that flows straight into a manufacturer's ERP system.
The claim is bold. Smartbase says it can reduce the manual labor of order entry by 90 percent—a vendor-stated figure that remains unverified by independent sources.
Whether manufacturers will believe that number—or whether it even matters if the product works well enough—is the kind of question that defines early-stage enterprise software.
The Problem That Won't Die
Manual order entry sounds quaint, almost anachronistic in an era when logistics giants optimize delivery routes with real-time satellite data. Yet talk to anyone who runs a small manufacturing operation and they'll tell you: purchase orders still arrive in bewildering variety. PDFs attached to emails. Faxes (yes, still). Occasionally, actual paper forms filled out by hand and scanned into oblivion.
Each one has to be keyed into an ERP system—SAP, NetSuite, Microsoft Dynamics, whatever the shop is running. The work is tedious, error-prone, and expensive. Even modest-sized manufacturers sometimes employ multiple full-time staff just to handle data entry. It's the kind of bottleneck that hasn't meaningfully changed in decades, despite waves of enterprise software promising to streamline everything.
Smartbase's workflow is straightforward, at least in theory. Customers upload scanned orders or connect an email inbox. The software extracts the relevant data using what the company describes as customizable business rules. A human reviewer checks and approves the parsed information. Then the order moves directly into the ERP.
The company's website lists integrations with platforms including Acumatica, Epicor, and Microsoft Dynamics 365, though the precise operational status of each integration isn't specified. Once approved, orders can trigger the printing of shop travelers directly from the ERP, a detail that suggests the founders have spent time understanding how metal shops actually operate.
A Familiar Playbook
Goldman, the CEO, isn't new to startups. He previously founded a DevOps platform that, according to the company's Y Combinator profile, raised $2 million before this venture. His co-founder, Fujioka, has a more corporate pedigree: the youngest SDE 2 at AWS SageMaker Inference, according to their Y Combinator profile, where he built AI infrastructure for enterprise customers including Salesforce.
A team of two. Targeting a narrow vertical. Hunting for design partners to validate the product before scaling. It's classic Y Combinator strategy, the kind of playbook that's produced both breakout successes and quiet shutdowns.
The company appears to be pursuing what it calls "five-figure annual contracts" with its initial customers, concentrated in metal finishing shops. Pricing information isn't publicly available, typical for startups at this stage. No case studies or customer logos have been published yet—standard for an early-stage company—though this does leave the 90 percent automation claim unverified by anyone outside the company.
Smartbase's own ROI calculator—every SaaS startup has one—projects annual savings of $81,000 and 1.4 headcount reductions per customer. Vendor math, in other words. Useful for framing a pitch, less so for making purchasing decisions.
Big Promises, Small Team
Here's where things get interesting, or perhaps concerning depending on your perspective.
Smartbase's website makes an unusually ambitious commitment for such an early-stage operation: plans for a dedicated U.S.-hosted enclave for customers subject to ITAR, CMMC, and DFARS requirements. The company even mentions plans for FedRAMP compliance, the federal government's cloud security standard.
Defense contractors and manufacturers handling sensitive materials represent a potentially lucrative market. They also demand serious infrastructure, extensive documentation, and the kind of operational maturity that typically requires more than two people. As of this writing, these compliance plans remain aspirational, with no public certifications or authority-to-operate documents—not surprising, perhaps, but worth noting for anyone evaluating the timeline on those compliance promises.
Whether Goldman and Fujioka can deliver that infrastructure remains an open question. Building secure enclaves and navigating federal compliance frameworks is a different challenge entirely from writing elegant code.
Not Exactly a Blue Ocean
Automated order entry isn't a new idea, and Smartbase isn't entering virgin territory.
Established vendors like Conexiom, Esker, and IntelliChief have offered AI-powered order capture for years. ERP companies themselves have started baking automation directly into their platforms—Epicor, for instance, has native modules designed to handle exactly this workflow. Meanwhile, newer startups like Arzana and Korso are attacking adjacent problems in manufacturing: AI agents for quoting, order coordination, supplier management.
What Smartbase does have is focus. By narrowing in on metal finishing shops and emphasizing seamless integrations, the company may find traction with job shops that lack dedicated IT staff and can't afford lengthy implementation cycles. These are businesses that need something that works out of the box, or close to it.
The sales-led approach—no self-serve tiers, just direct conversations leading to those five-figure contracts—suggests the founders understand they're selling to skeptical buyers who've heard plenty of software promises before.
The Waiting Game
For now, Smartbase is doing what vertical SaaS startups do at this stage: articulating a clear value proposition, securing early customers, and racing to build the product and compliance infrastructure needed to justify expansion.
The 90 percent time reduction sounds great in a pitch deck. Whether it holds up after six months of real-world use—complete with edge cases, unusual order formats, and the inevitable integration hiccups—will determine if manufacturers actually write checks, or if Smartbase becomes another well-intentioned tool that overpromised and underdelivered.
The founders have targeted what appears to be a persistent pain point in manufacturing workflows, which remain stubbornly manual in pockets where you'd least expect it. Whether two engineers and an AI model can unseat entrenched habits and legacy systems is a different matter entirely.
The answer, as with most enterprise software bets, won't be clear for a while yet.
