JAGGAER vs Brex vs Medius for AP Automation

For a 3-person AP team at a $120M multi-location services company currently managing payments across disconnected bank portals, Brex Bill Pay consolidates all four payment rails into a single dashboard. Within the Bill Pay module, users upload invoices, route them through approval flows, and then pay via ACH, domestic or international wire, check, or Brex card, with the result synced to the connected ERP. The product supports scheduling one-time or recurring ACH payments, checks, or wires, including international wires in foreign currency, from any connected bank account, with Brex's enhancements having opened up access to third-party bank accounts for greater funding flexibility. Virtual card payments are also supported: the AP user selects 'pay by one-time virtual card' on a bill, Brex generates a card available to the designated cardholder on the scheduled payment date within the dashboard, and the buyer earns Brex rewards points on that vendor spend. This covers the payment execution stage of the AP pre-processing journey (post-approval, post-matching), operating as a straight-through payment hub after invoice coding and approval steps are completed upstream in the same Brex Bill Pay workflow.

For a $120M multi-location services company currently running fragmented bi-weekly check runs and monthly ACH batches through manual bank portals, Medius Payments consolidates all disbursement rails into a single interface. The module, branded as Medius Payments (built on the acquisition of OnPay Solutions and launched as an evolution of Medius Pay), covers all four rails the buyer requires: ACH, check, wire, and virtual card for US-based operations are handled in a single consolidated process for each supplier, with no need to log into a separate bank portal for ACH transmission. ACH is issued as Electronic Funds Transfer via Automated Clearing House in a secure, controlled environment, with no need to log into a bank or other website to transmit payments. Wire transfers are formatted and transmitted from within the platform, with support for SWIFT and IBAN, formatted and sent via secure FTP to the financial institution's wire desk. For virtual cards, Medius manages a supplier enrollment campaign for vCards, with more than 25% of vendors already accepting them. Payments sync automatically back to the ERP, closing the loop from invoice approval to ERP posting without a separate reconciliation step. This operates at the final stage of the pre-processing journey (stage 5: payment execution), downstream of the invoice capture, matching, and approval stages Medius also covers.

For a $120M multi-location services company moving off manual email-and-check AP, JAGGAER Pay is the payment execution module embedded within the JAGGAER One source-to-pay suite. Once invoices are approved, JAGGAER Pay can automatically create, batch, and process payments from a single interface, eliminating the need to log into separate banking portals. ACH and virtual card are delivered via two embedded fintech partnerships: Finexio (active since 2022) and Bottomline Paymode (added September 2024), with the latter covering Premium ACH and virtual card on a 600,000-supplier network. Wire transfer is handled through a separate embedded partnership, PaymentsHub by TransferMate, which covers ACH, wire, and instant payments across 200+ countries from within the JAGGAER interface. Printed check capability is documented by Finexio as part of its JAGGAER Pay stack. The JAGGAER homepage explicitly states that 'with JAGGAER Pay, every AP task [is] managed in a single solution, eliminating the need to bounce between ERPs and multiple-point solutions.' The payment hub operates at the tail end of the pre-processing journey (post-approval, post-matching) and does not replace earlier stages such as receipt confirmation or approval routing. The material limitation for this buyer is that the four rails are assembled across multiple embedded fintech partners (Finexio, Paymode, TransferMate) rather than a single native payment rail; the 'single interface' claim is real, but the underlying architecture depends on which partner modules are active in the buyer's specific subscription tier, and JAGGAER's platform is primarily sized and priced for large enterprise procurement organizations, not $120M mid-market services companies.

For a multi-location services company processing 1,800 invoices per month across a varied vendor mix of facilities, subcontractors, utilities, and subscriptions, Medius addresses this requirement through two named, integrated modules: Medius Capture and SmartFlow. Medius Capture uses OCR, AI, and machine learning to extract header and line-item data from paper and digital invoice formats; Medius Capture has been refining its AI and Machine Learning algorithms since 2016, learning from hundreds of millions of invoices to deliver 100% touchless capture rate. The learning mechanism operates at the per-vendor level via a confidence-scoring threshold: highly accurate and reliable data capture that learns after just two invoices allows for Touchless Capture to be widely enabled, so invoices with very high levels of data confidence simply bypass manual verification and fly through for processing, where SmartFlow takes over to automatically code and distribute invoices. The feedback loop that drives continuous improvement is explicit: AI algorithms learn from every invoice processed, increasing in accuracy and efficiency. For the buyer's non-PO invoices (utilities, subscriptions, professional services), SmartFlow handles the GL coding side of the learning curve: Medius AP Automation features SmartFlow, a powerful AI model that auto-fills coding, tax, and approver values for non-PO invoices with 95% precision after just two invoices, bringing about unheard-of efficiency and powering the industry's highest rates of straight-through processing. The system's agentic AI adapts as supplier formats evolve without manual template maintenance: machine learning models can recognize and adapt to different invoice formats, improving over time with every document processed; instead of relying on fixed templates, AI interprets context and patterns to identify key information accurately, and with Agentic AI, the system can autonomously adjust to new invoice formats or anomalies without human input. Exceptions that do require human correction feed back into the model: as the system learns, the number of exceptions continues to fall, and AP teams spend more time on high-value work. This operates at Stage 1 of the pre-processing journey (legitimacy and data extraction) and extends into Stage 5 (cost allocation via automated GL coding), but does not replace the human steps required for Stage 3 (contract terms verification) or Stage 4 (receipt confirmation for 3-way matching).

For a 6-location services company processing 1,800 invoices per month across a mixed PO and non-PO book, JAGGAER offers two distinct layers of invoice capture AI with meaningfully different learning depths. The native Digital Capture module, introduced in release 21.1, combines OCR with machine learning and is described as enabling 'no-touch invoice capture via machine learning' with the ability to 'set confidence thresholds, review key decision factors, and continuously refine AI accuracy' -- but published documentation does not describe a per-supplier adaptive learning mechanism, such as corrections feeding back into a vendor-level recognition profile or supplier-format-specific model retraining. The more substantive learning capability sits in the AppZen Autonomous AP integration, available as a connected layer within JAGGAER Invoicing: its own product page states it 'learns from your feedback without further training' and a live RPI deployment achieved 97% data capture accuracy on 12,000+ invoices processed in six months. However, AppZen is a third-party partner connected via JAGGAER's REST API -- it is not part of the base Digital Capture module -- meaning the buyer must adopt and configure an additional component to access the documented feedback-learning loop. At stage 1 (legitimacy) and moving toward stage 2 (PO match), both layers operate on incoming invoice data before the invoice reaches the ERP; neither substitutes for receipt confirmation at stage 4.

For a 3-person AP team processing 1,800 invoices per month, Brex Bill Pay operates at stage 1 of the pre-processing journey (invoice capture and data extraction). When an invoice arrives via email forward, drag-and-drop upload, or vendor upload link, Brex AI technology immediately captures data from PDFs and email attachments and automatically extracts key information like amounts, due dates, and vendor details. The underlying extraction engine is LLM-based: Brex evolved its invoice scanning from traditional OCR technology with 80% accuracy to advanced LLM-powered technology that delivers 97% accuracy and blazing-fast line itemization. The LLM is not proprietary: Extend fine-tuned dedicated models for Brex's workloads and deployed them on private GPU infrastructure, avoiding latency spikes and resulting in the highest accuracy Brex had ever seen, with many document tasks reaching 99%. On the learning question specifically, Brex's marketing pages describe correction-driven improvement: the platform learns from corrections and gets smarter over time, and even handles complex multi-page invoices with tables and various formats. However, the actual help center documentation describes a different mechanism: when an invoice is uploaded, Brex uses AI to pre-fill the text from the invoice to the fields, and if AI reads these fields incorrectly, users can manually adjust them. There is no help-center documentation of a supervised feedback loop that routes AP user corrections back into a vendor-specific or customer-specific retraining pipeline. The Extend architecture indicates a shared, platform-level fine-tuned model rather than per-customer model adaptation.

For this $120M multi-location services company processing 55% PO-based invoices across two Sage Intacct entities, JAGGAER's native matching engine covers all five stages of the pre-processing journey for PO-backed transactions. JAGGAER's AP module applies configurable 2-, 3-, and n-way matching rules to POs, receipts, and more, with built-in alerts and collaboration tools that streamline exception handling to keep processes moving efficiently. The tolerance configuration is buyer-administered: matching configuration options including rules and tolerances are configured at AP Administration > Matching Rules and Tolerances pages. The system supports percentage-based thresholds, as documented in the official help center: an organization may set tolerances of 3% above the cost and 10% below the cost, and whenever documents meet these criteria, the invoice is automatically matched and marked as payable, confirming the buyer's specific 2% price and 5% quantity thresholds are configurable within the same framework. Receipt confirmation (Stage 4 of the pre-processing journey) is natively integrated: Receipt Lead Time defines the number of days required for a shipment to be received and the receipt to be entered into the system; the auto-match robot holds unmatched invoices that are missing receipts until the receipt has been entered and the invoice has been matched. Invoices that fall outside configured tolerance bands are routed to an exception queue rather than auto-approved: the system categorizes PO/Invoice/Receipt matches as Within Tolerance, Outside of Tolerance, or Do Not Match, enabling the AP team to review only true exceptions. AI automatically matches invoices to purchase orders and goods receipts even when formats vary, evaluates tolerance thresholds, historical patterns, and supplier behavior to determine whether an invoice can proceed or requires review, and where exceptions do occur, AI-supported exception handling helps teams focus on what genuinely needs attention.

For this buyer's 55% PO-based invoice volume covering facilities, supplies, and subcontractors, Medius executes automated three-way matching by synchronizing vendor ledgers, purchase order details, and goods receipts from Sage Intacct into the matching engine, then comparing the inbound invoice against all three documents before routing. To enable automated 3-way matching, Medius requires that master data including vendor ledgers, purchase order details, and goods receipts synchronize seamlessly between the ERP system and the AP automation solution. When all data aligns within applied tolerance levels, the invoice flows touchless to ERP posting without manual review. 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 buyer's specific 2% price and 5% quantity thresholds are directly configurable via Medius's connection tolerance feature: an admin or AP user can specify an acceptable range in amount or percentage for automatic connection at the header or line level, this can be done at the company and supplier levels, with separate limits settable for positive and negative deviations, allowing automatic connection between invoice and purchase orders or goods receipts as long as amounts are within the defined threshold. Invoices that fall outside tolerance are classified as deviations and surfaced for exception routing, covering both pre-processing journey stage 2 (PO match) and stage 4 (receipt confirmation). Automated systems allow for flexible tolerance levels, ensuring that minor discrepancies in quantities or pricing are automatically processed without delay.

For a $120M services company with 55% PO-based invoices covering facilities, supplies, and subcontractors, the matching ceiling matters directly for roughly 990 invoices per month. Brex's bill pay module performs automated two-way PO matching: when an invoice arrives, the AI scans it and attempts to match it to an open PO synced from the ERP, surfacing the PO number, date, and remaining available amount for AP review; if no automatic match is found, it presents the open POs for that vendor for manual selection. Brex AI matches the imported invoice to an open PO in the ERP "via our two-way accounting integrations," framing the mechanism explicitly as two-way. Brex customers "rely on PO matching to help prevent fraud and automate compliance"; if no match is found automatically, Brex displays open POs for manual selection. This covers Stage 2 of the pre-processing journey (PO match) but stops there. No product documentation, help center article, or feature page found in any search confirms that Brex's bill pay module incorporates a goods receipt confirmation step as a third document in the match, meaning Stage 4 (receipt confirmation: did we actually receive the goods?) is not handled by the system. Brex's own AP automation content describes the automated check as two-way matching that "automatically compares invoice details against purchase orders to verify accuracy," checking quantities, prices, and terms, and flagging discrepancies for review. Separately, no source documents that Brex exposes buyer-configurable tolerance thresholds (such as the buyer's required 2% price, 5% quantity bands) that determine whether a variance auto-approves or routes to an exception queue.

This $120M multi-location services company needs assurance that Sage Intacct integration setup will be handled during implementation without a separate statement of work or added cost. JAGGAER does provide a Professional Services team and a structured implementation methodology covering deployment, project governance, technical consulting, and integration design; and its JAGGAER Link offering includes prebuilt ERP connectors for Oracle, SAP, NetSuite, and Ellucian. However, Sage Intacct is not listed among the named prebuilt connectors in JAGGAER's Connect/Link product. Integration with Sage Intacct would rely on JAGGAER's public REST API framework, under which the client bears responsibility for building, testing, and staffing the integration, while JAGGAER Professional Services provides 'guidance and support.' JAGGAER's own integration documentation explicitly lists client responsibilities that include building and developing the integration, transmitting files, staffing 'knowledgeable technical resources,' and migrating client-side configurations, which means integration setup is not a bundled zero-cost deliverable. Furthermore, third-party pricing sources confirm that 'implementation and integration services may incur additional costs, depending on the complexity of your existing systems,' which directly contradicts the buyer's requirement that integration setup be included without a separate SOW.

For a $120M, 2-entity Sage Intacct customer running 1,800 invoices per month, Brex offers a native, self-serve Sage Intacct connector that Brex's own documentation describes as one that 'does not require any additional cost,' with setup completed directly inside the Brex dashboard by enabling Web Services in Intacct, creating a Web Services User, selecting entities, and mapping GL accounts, departments, and custom dimensions. The help center publishes the full configuration walkthrough, including entity selection and field mapping steps, positioning this as a guided self-setup rather than a professional services engagement. At the Enterprise tier, Brex explicitly bundles 'comprehensive professional services, change management, and ongoing support' to reach go-live in 6 to 8 weeks, framing implementation assistance as included rather than a separate SOW. However, the self-serve documentation is oriented primarily toward expense and card transaction sync; the bill pay and AP invoice sync path for Sage Intacct ends with a 'Talk to our sales team to learn more' call-to-action rather than a published configuration guide, signaling that full AP integration depth may require sales-negotiated onboarding support. Third-party procurement intelligence (Vendr) further notes that 'complex implementations, such as multi-entity accounting integrations,' may incur separate professional services fees ranging from $5,000 to $50,000 or more, depending on scope, even though these fees are often negotiable.

This buyer runs 2 Sage Intacct entities and needs the connector configuration, field mapping, and go-live setup included in the base implementation fee rather than billed as a separate SOW. Medius does offer a pre-packaged Sage Intacct integration, but the architecture here is materially different from its flagship ERP connectors: Medius has a pre-packaged integration with Sage X3 and Intacct in partnership with Acuity Solutions, a UK-based Sage specialist, rather than through Medius's own 'fully managed' connector infrastructure that covers SAP, Oracle, NetSuite, Infor, and Microsoft Dynamics. Medius has fully managed integrations to leading ERP solutions from Microsoft, Oracle NetSuite, Oracle JDE, Oracle Fusion, Infor and SAP, but Sage Intacct is not on that list: it sits in the partner-delivered tier. For its first-party ERP connectors, Medius markets a Rapid Application Deployment (RAD) methodology: pairing pre-packaged ERP integrations with a proven RAD methodology gives finance teams a streamlined path to automation without complexity or delays, with a best-practice approach that minimizes disruption and reduces IT lift. The homepage also states "You won't have to worry that lengthy implementation or ongoing administration will add consultants and costs" with average go-live in 8-12 weeks, but this is a general marketing statement covering all implementations, not a contractual commitment specific to the Sage Intacct partner path. Because the Intacct connector flows through a third-party partner engagement, there is no public evidence that connector configuration, dimension mapping, and entity setup are bundled within Medius's standard implementation fee rather than scoped and billed separately by Acuity Solutions.