Somewhere on a shop floor right now, a supervisor is writing down what he needs on a piece of paper.
Safety glasses. A bolt for the machine that's been down since Tuesday. A replacement weld gun.
He hands the paper to a purchaser. The purchaser looks at it and immediately has questions. What spec on the bolt? Which lens grade on the safety glasses? Is this the same weld gun as the one replaced six months ago, or a different model? Is this budgeted or does it need a manager sign-off first?
The supervisor is already back on the floor. The purchaser leaves a voicemail. The bolt doesn't get ordered until Thursday. The machine stays down.
This is the purchasing requisition process at most mid-market manufacturers. And it is costing far more than most operations leaders realize.
The Hidden Cost of Human Friction
The paper-based PREQ is not just slow. It's structurally designed to fail.
Here's what actually happens when a manual requisition process meets a complex manufacturing environment:
The information is incomplete
The supervisor knows what he needs but not necessarily how to specify it in a way a purchaser can act on. Part numbers, quantities, required-by dates, approved vendors, specification documents. None of this is prompted. It has to be chased down after the fact, which adds days to the cycle time on purchases that were already urgent.
The timing is always late
By the time a manual PREQ reaches purchasing, the need is usually already behind schedule. The machine is already down. The safety stock is already depleted. Manual processes create reactive purchasing and reactive purchasing is expensive purchasing.
Approvals are invisible
Who authorized this? Is it within budget? Does a spend threshold require manager sign-off? In a paper process, these questions either create bottlenecks at every transaction or get ignored entirely, creating compliance exposure that shows up later in audits or budget variances.
Vendor selection is tribal knowledge
Without a structured process, purchasers fall back on whoever they already know. Preferred vendor lists go unused. Volume gets scattered across one-off suppliers. The leverage that comes from consolidated, managed spend never materializes for the categories that actually drive the most transaction volume.
The aggregate effect: your highest-frequency purchasing category is also your least managed one, running on a process that was designed for a different era and has never been systematically improved.
Where AI and Automation Change the Math
This is a problem that structured software and AI are genuinely well-positioned to solve, but not in the way most companies approach it.
The instinct at most organizations is to "get the PREQ process into the ERP system." And in theory, that's correct. A centralized system should be the backbone of any procurement workflow.
The problem is that most legacy ERP systems were not designed with the flexibility to handle the exceptions, thresholds, and conditional routing that a real MRO purchasing process requires. Getting a PREQ workflow fully configured in a complex ERP, with appropriate approval tiers, vendor routing logic, spend thresholds, and category-specific fields, is the kind of project that takes months longer than anyone planned and often never fully gets implemented. The complexity is real. The customization requirements are extensive. And in the meantime, the shop floor keeps running on paper.
The smarter path is to separate the intake and workflow logic from the ERP, at least initially, and let purpose-built tools do what they're actually designed to do.
Structured digital intake
An AI-guided PREQ form prompts the requestor (the supervisor, the maintenance tech, or whoever is initiating the need) through the information a purchaser actually requires to act. Part number or description. Quantity. Required date. Suggested vendor if known. Relevant specification or equipment. This isn't complicated technology. But it eliminates the back-and-forth that makes manual PREQs slow.
Spend threshold routing
Below a defined dollar threshold the requisition is auto-approved and the purchaser gets a clean, complete order to execute. Above the threshold, the system routes the request through the appropriate approval workflow before it reaches purchasing. No paper. No chasing. No bottlenecks that require a manager to physically sign something before a $75 part can be ordered.
Catalog and preferred vendor integration
For commonly purchased MRO items, like consumables, standard hardware, safety supplies, an AI-powered catalog layer presents the requestor with approved items at pre-negotiated prices before they even reach the purchasing team. The supervisor searching for safety glasses sees the approved SKU, selects it, and submits. The purchase executes against a managed supplier relationship at a known cost. This is the version of the PREQ process where purchasing becomes largely invisible not because it's been eliminated, but because it's been embedded intelligently upstream.
Automated purchase-to-pay execution
Once the PREQ is clean, approved, and routed correctly, the downstream steps of PO generation, vendor communication, receipt confirmation, invoice matching, payment can be substantially automated. The purchaser's role shifts from processing transactions to managing exceptions. The exception rate drops as the intake quality improves. The whole cycle time compresses.
The Trap: Digitizing a Broken Process
One caution worth naming directly.
If the PREQ process itself is poorly designed, for example if the approval tiers don't reflect how the business actually operates, if the vendor routing doesn't align to real supplier capabilities, if the spend thresholds haven't been thought through then digitizing it doesn't fix it. It just makes the dysfunction faster and harder to see.
This is why the foundation work matters before any technology is deployed. What does a complete, actionable PREQ actually look like for your environment? What are the real approval thresholds that reflect your risk tolerance and budget structure? Which MRO categories are candidates for catalog management and which require human judgment on every transaction?
Getting those questions right is the prerequisite. The technology is the accelerant. But you can't accelerate a process that hasn't been correctly designed.
What Good Looks Like
A well-designed, AI-supported PREQ process for MRO and miscellaneous purchases does a few specific things:
It forces completeness at intake so purchasing receives requests that are actionable, not questions waiting to be answered. It routes intelligently based on spend, category, and approval authority so the right person sees the right request at the right time. It connects to preferred vendors and manages catalogs for high-frequency items so low-value transactions don't consume disproportionate purchasing bandwidth. And it feeds clean, structured data into the broader spend analytics picture so the tail spend visibility we talked about elsewhere actually materializes.
The result is a procurement team that spends its time on decisions, not administration. And a shop floor that gets what it needs faster, with less friction, and within the guardrails the business actually intends.
That's not a technology transformation. It's a process improvement with technology as the enabler.
And for most manufacturers we work with, it's one of the highest-return changes available precisely because the current baseline is so low.