RFP Requirements: AP Automation (Manufacturing) ## PO: Comparison

This manufacturing buyer needs automated 3-way matching (PO + GRN + invoice) with configurable tolerance thresholds segmented by commodity type, vendor, plant, or dollar amount against their NetSuite ERP. Medius delivers this through its PO Invoice workflow, which operates at pre-processing stage 4: it ingests goods receipt data alongside PO and invoice data, compares all three at the line level, and enforces configurable deviation tolerances before any human touch occurs. The Medius-NetSuite integration (a dedicated managed cloud connector) synchronizes PO lines, goods receipt records, and vendor ledgers directly from NetSuite into the Medius matching engine; the May 2025 NetSuite-specific product sheet confirms that Medius imports unit price and quantity from PO line details for the most accurate match, then falls back to line total or header amount matching for service-based lines. Tolerance configuration is an explicit, admin-managed feature: Medius allows AP teams to set percentage or fixed-amount deviation thresholds per supplier, and the customer success documentation confirms tolerances can also be set by the nature of the goods or services a specific supplier provides, which maps to commodity-type segmentation. When all invoice lines fall within configured tolerances against PO and GR data, the invoice proceeds to automatic touchless posting in NetSuite with no human review; when a missing GR is detected, Medius routes the invoice to the responsible user first and holds any price-deviation routing until the GR is recorded, enforcing true 3-way match discipline.

For a manufacturing buyer running NetSuite and needing 3-way matching with commodity-differentiated tolerance rules, Stampli's Billy AI connects POs, goods receipts, and invoices in real time, performing both 2- and 3-way matching before any manual review occurs. Billy connects POs, receipts, and invoices in real time, performs 2- and 3-way matching, notifies teams when items are received or missing, and keeps ERP records in sync. On the tolerance side, Stampli automatically skips invoice approvals when POs and invoices match based on customer-defined tolerances, with automatic identification of exact matches and discrepancies tied to those tolerance settings. Documented tolerance dimensions include vendor-specific rules and dollar-amount thresholds: a buyer can establish a rule allowing a 5% price difference from specific vendors while flagging all others, and can configure the system to automatically verify high-value invoices as the most fraud-prone. Billy also links invoices to the correct POs or receipts at the line level before any human action: Billy codes invoices line by line, validates vendors and required fields, flags duplicates, and links invoices to the right POs or receipts, all before anyone lifts a finger. The gap is commodity-type-level tolerance configuration: Stampli's product pages and help center document tolerance rules by vendor and dollar amount, but no published documentation confirms that tolerance thresholds can be pre-configured as named rule sets keyed to a commodity category or item class (e.g., raw steel at +/-2%, precision-machined at exact match). That dimension of tolerance hierarchy is present natively in NetSuite's own 3-way match workflow at the item and vendor record level, but Stampli's layer does not explicitly surface commodity-type tolerance as a standalone configuration axis within its own UI.

For a manufacturer needing four distinct commodity-category tolerance bands (raw steel at +/-2% quantity, precision components at exact match, MRO at +/-5%, hazardous chemicals at exact match for regulatory tracking), Stampli's PO matching engine sits at stage 2-3 of the pre-processing journey: it performs automated 2- and 3-way matching at the line level and supports customer-defined tolerance settings that can trigger automatic approval bypass when invoices fall within range. As documented on Stampli's PO Matching product page, the system features 'automatic identification of exact matches and discrepancies, coupled with customer-defined tolerance settings,' and the AI Line-Level PO Matching page confirms tolerances drive touchless approval: 'automatically skip invoice approvals if POs and invoices match based on customer-defined tolerances.' A real NetSuite customer (Ollie, a food manufacturer) confirms that Stampli's 'acceptable ranges' suppressed re-approval for inventory invoice variances. However, every documented and customer-confirmed instance of tolerance configuration operates at a global or vendor-level scope, not at a commodity-category or item-type level. No Stampli documentation, help article, or case study surfaces a mechanism to assign distinct tolerance percentages to separate item categories (e.g., raw steel vs. MRO vs. hazardous), to enforce UoM-aware quantity tolerance logic for weight-based materials specifically, or to flag a commodity class (hazardous chemicals) for mandatory exact-match enforcement tied to a regulatory classification. The tolerance engine that exists is real and functional; it simply lacks the per-commodity-class differentiation this buyer requires.

For a manufacturing buyer needing differentiated tolerance rules per commodity category (raw steel at +/-2%, precision-machined components at exact match, MRO at +/-5%, hazardous chemicals at exact match), Medius operates at pre-processing stages 2 and 3: PO connection and deviation validation. Its deviation tolerance engine is well-documented and supports quantity tolerance, unit-price tolerance, line-total tolerance, and header-amount tolerance, each configurable by absolute amount or percentage with separate positive and negative limits. However, the documented scoping levels for these rules are company-level and supplier-level only. Business rules are defined at the company or supplier level, covering five deviation types configurable by amount or percentage: header amount, line total, unit price, and quantity. Supplier-level overrides allow per-vendor fine-tuning: in Medius it is possible to configure which connection types are used on each vendor; supplier-specific match settings override the company configuration. A third-party analyst review also notes Medius has over 500 out-of-the-box business rules for matching and validations, which can be turned on or off and configured with specific tolerances and exception values, and that rules include head- and line-level tolerances, price unit tolerances, and quantity tolerances, as well as validation of account codes and distribution costs. No source in official Medius documentation or help center articles confirms that tolerance rules can be scoped to a commodity category, item class, or material group dimension. The buyer's requirement demands exactly that: four different tolerance bands applied to four different material types, some of which may arrive on the same invoice from the same supplier. Supplier-level rules cannot differentiate between raw steel and precision-machined components invoiced by the same vendor.

For a manufacturing buyer running PO-based procurement through NetSuite, Medius delivers touchless auto-approval as a named, configurable workflow outcome: when an invoice clears 3-way matching (PO, goods receipt, and invoice) within configured tolerance thresholds, automated invoice matching enables fully touchless invoice processing -- when all data on the invoice matches supporting documents and applied tolerance levels, the invoice can go directly from receipt and data capture to final posting in the ERP for payment without anyone reviewing or touching it. The mechanism operates through two coordinated layers. First, the "Touchless Capture" module uses AI and OCR to extract invoice data with high confidence, so invoices with very high levels of data confidence simply bypass manual verification and fly through for processing. Second, the matching engine applies tolerance rules and, for invoices that pass, triggers straight-through processing: Medius supports straight-through invoice processing, allowing invoices to flow from receipt to payment with no manual work, reducing errors and speeding up cycle times. Tolerance-based auto-approval of deviations is explicitly documented: tolerance levels can be set for fees like freight to avoid PO invoices getting stuck in the matching process, and an AP automation solution can automatically approve certain deviations during the matching process following preset rules. The system tracks performance through a built-in "Touchless Ratio" KPI, with Medius claiming a 96.3% touchless processing rate for PO invoices versus a 23.4% market average (Ardent Partners, 2025). This covers pre-processing stages 2 (PO match) and 3 (terms/tolerance check) and -- critically -- stage 4 (receipt confirmation via GRN), since upon delivery of goods or services, the buyer registers a goods receipt, and since approval of the cost is gained at the purchase requisition phase, the invoice does not need another round of approval as long as invoice details match the PO and goods receipt.

For a manufacturing buyer processing high volumes of PO-backed invoices across multiple commodity types, Stampli does support configurable auto-approval for matched invoices: its own documentation states that 'AP solutions like Stampli can be configured to skip the approval stage for invoices that match the associated purchase order exactly,' and that 'skipping approvals speeds up invoice processing and payment times.' Billy the Bot completes the pre-processing stage by linking invoices to POs and receipts before any human involvement, and the system can be configured to bypass the approval queue when matching criteria are satisfied. However, the mechanism Stampli documents most clearly is exact-match auto-approval, and Stampli's own positioning blog concedes that auto-approval is straightforward only for simple, single-line, verified-vendor scenarios. The commodity-specific tolerance band configuration this manufacturing RFP requires -- different tolerance thresholds by commodity category, plant, or vendor driving touchless routing -- is not documented at the rules-engine or configuration level in Stampli's help center, leaving a material gap between what is confirmed and the granularity of tolerance-driven straight-through processing the buyer needs.

For a manufacturing buyer running multiple production jobs, work orders, and cost centers through NetSuite, Stampli's AI Line-Level PO Matching module operates squarely at stages 2 and 3 of the pre-processing journey (PO matching and terms verification), with stage 4 (receipt confirmation) covered via live item-receipt sync from NetSuite. On the core multi-line, multi-allocation scenario: Stampli supports one-PO-to-many-invoice and many-PO-to-one-invoice scenarios, plus true 3-way matching at the item receipt level, with live PO and receiving data refreshing every two hours or on-demand to validate against actual received quantities rather than just PO headers. The NetSuite integration carries full allocation dimensions into Stampli: Stampli automatically mirrors any header or line-level custom field and can even map saved-search results into those fields, automapping any new custom fields so only relevant fields post back to the ERP without re-engineering. NetSuite-specific synced dimensions explicitly include subsidiaries, project, department, and warehouse fields. On structural mismatches where vendor invoice grouping differs from PO line structure: Billy learned to take the language on vendor invoices and translate it into internal language, handling different SKU names than what vendors use and picking up on changes in vendor invoice language, with one NetSuite customer (Ollie, a food manufacturer) reporting that breaking down flat poundage into individual SKUs dropped from two to four hours per day to minutes. When AI cannot fully resolve a structural mismatch, users can add or adjust line items within the items or GL table to ensure accuracy, and resolve discrepancies with tracked questions and responses directly on the invoice processing page. Billy codes invoices line by line, applying GL accounts, departments, and custom dimensions learned from payment and accounting history, validating vendors and required fields, flagging duplicates, and linking invoices to the right POs or receipts before anyone lifts a finger. The material ceiling is the split-line scenario: when a single vendor invoice line must be disaggregated across multiple PO lines each pointing to a different job, work order, or cost center, Stampli surfaces this through its line-level adjustment UI and GL table with auto-calculators, but fully automated cost-center-specific split attribution (one invoice line mapped to PO line 1 at Job A and PO line 2 at Job B without any human confirmation) is not explicitly documented as a zero-touch automated function.

For a manufacturer routing multi-line PO invoices across production jobs, work orders, cost centers, and projects in NetSuite, Medius operates a tiered line-level matching engine: it first attempts a detail-level connection matching invoice line item numbers and descriptions to PO line details (unit price and quantity validation), then falls back to a Line Total Match comparing net amounts per invoice line against PO/delivery line net amounts, and finally applies a Line Amount Connection for service-based lines. Medius can 'accurately captur[e] information from invoice headers and line-items then automatically matching them against single or multiple POs and receivables,' and the platform explicitly supports 'Multi-invoice to Multi-POs and PO lines matching,' connecting the right information from the invoice, purchase order line, contract and goods receipt automatically. NetSuite coding dimensions are pulled in at implementation: the structure of coding dimensions, including both standard and custom dimensions, is determined during the data-gathering phase of onboarding, and Medius imports all coding dimensions directly from Oracle NetSuite. However, the documented mechanism reveals a material ceiling for this buyer's scenario: if discrepancies exist, such as incorrect product codes, the invoice is halted for manual connection, and for quantity differences, no automatic coding is applied; instead, businesses must resolve these variations manually by addressing delivery discrepancies. When a vendor consolidates or splits line items differently than the PO structure, the structural mismatch surfaces as a failed automatic connection that drops into a manual resolution queue rather than being autonomously remapped to the correct PO line and cost allocation.