Medius vs Spendesk vs Esker for AP Automation

This $120M multi-location services company needs a forward-looking payables view that shows approved and pending invoices bucketed by due date, so the AP team can project cash outflows before each bi-weekly check run or monthly ACH batch. Medius addresses this through its Medius Analytics module, which is described as an always-on spend analysis solution. Medius Analytics is positioned to help users "never lose track of your cash position" through a spend analysis view of cashflow that also surfaces opportunities to extend supplier payments to manage cash position. The AP automation product page describes the capability as the ability to "stay on top of costs, forecasts and cashflow with a full, real-time view into supplier invoices through pre-defined reports, dashboards and KPIs." The platform also claims that "with insight into direct spend, indirect spend, and employee expenses, you'll know precisely what's pending and what's paid, building trust in forecasts and eliminating surprise expenses." At the blog level, Medius describes AP automation software that can "provide analytical tools that forecast future cash flow based on historical data and pending transactions." Specifically, "Medius Analytics can monitor these factors continuously, adjusting forecasts in real time and providing predictive insight into how vendor activity or pricing changes will impact cash flow." The ceiling, however, is that no Medius help documentation or product page reviewed confirms a discrete, named report that explicitly buckets projected outflows by forward-looking due-date horizon (e.g., 7/14/30/60/90-day bands) with a status dimension distinguishing invoices pending approval from those approved and awaiting payment. The Medius gadget framework allows configurable dashboard panels that display real-time data, and the platform tracks open versus archived invoices, but the specific forward-looking due-date distribution view the buyer requires is described only at the marketing and blog level, not documented as a discrete named feature with a confirmed mechanism.

For a $120M multi-location services company currently running manual AP with no visibility into future cash obligations, Esker's AP analytics layer provides a named 'AP cashflow' dashboard widget and explicitly lists 'Payment forecast and prediction reports' as a P2P cycle metric within its platform. Esker's solution is equipped with intelligent dashboards that display live, visual analytics including 'AP cashflow, AP process metrics, spend by category, volume and supplier, requests pending approval, budget control and forecasts.' At the P2P cycle metrics layer, Esker's documented analytics include 'Payment forecast and prediction reports' alongside aging reports for AP invoices, DPO reports, and risk evaluations. The AP solution page further commits to this use case: Finance teams are given 'the visibility and control they need to manage spend, anticipate cashflow, fraud prevention and support compliance across the AP process.' The platform tracks due dates at the invoice level: Esker 'manages and analyses invoice data, automatically queuing received requisitions and tracking the important details mentioned above: PO numbers, due dates, early payment discounts, and so on.' However, the specific forward-looking mechanism on the payables side (explicit time-bucket distribution such as 7/14/30/60/90 day bands, with invoices segmented by approval status: pending vs. approved for payment) is documented at the AR collections side with named 30-90 day granularity but is described only at the KPI/dashboard-widget level on the AP side. No help center article or technical specification was found that confirms the AP cash flow forecast surfaces a due date distribution by approval status at the depth the buyer requires for treasury planning.

This $120M services company needs a forward-looking payables forecast that shows cash outflow requirements by due date across both pending (unapproved) and approved invoices. Spendesk's closest mechanism is the 'All Payables' page in the Bookkeep module, which surfaces a filterable ledger of payables including due date, paid status, bookkeeping status, and cost center fields available for CSV export. Per-invoice payment scheduling is also available at the controller review stage, allowing specific disbursement dates to be set on approved invoices. However, a critical data model constraint eliminates the buyer's pending-invoice requirement entirely: as documented in Spendesk's developer API reference, 'invoices and expense claims that have been submitted but haven't yet been approved do not have a payable created,' meaning pre-approval invoices are invisible to the payables layer and cannot appear in any payables report. The real-time reporting that does exist reflects budget consumption as invoices are approved, not a forward-looking distribution of outflows across future time periods. The Spendesk blog on cash flow forecasting is a general educational guide that directs finance teams to external tools for building forecasts, with no reference to a native forecasting module.

For a $120M multi-location services company running bi-weekly check runs through Bank of America, the critical question is whether Medius generates a formatted positive pay file that can be submitted directly to BoA's CashPro portal after each check run. Across four targeted searches of success.medius.com, medius.com product and payments pages, and general search terms combining 'Medius,' 'positive pay,' and 'Bank of America,' no documentation, help article, product page, or release note was found describing a positive pay file generation mechanism in any Medius module. The Medius Payments product describes check handling through an off-site print-and-mail facility and positions its fraud controls as internal AI-based anomaly detection before payment execution, not as a bank-side check-issuance file submission after a check run. The Medius blog on check fraud prevention explicitly advocates digital payment migration as the fraud control strategy rather than referencing positive pay file output as a product capability.

This $120M multi-location services company running bi-weekly check runs needs a positive pay file formatted to Bank of America's exact specification, exported as a post-payment deliverable from the AP platform. Esker Pay, Esker's payment automation module, supports check payments alongside ACH, wire, and virtual card, and the platform documents 'bank details verification and fraud detection' as a capability within its S2P payment workflow. However, that documented mechanism is supplier bank account verification at the point of onboarding, which validates that a supplier's account credentials are legitimate before payment is initiated. This is architecturally distinct from positive pay file generation, which produces a fixed-format check issuance file (check number, amount, date, payee) transmitted to the bank after a check run so the bank can match and validate presented checks. No Esker product page, datasheet, help article, or documentation found in four targeted searches references positive pay file generation, bank-specific check format templates, or Bank of America CashPro file output as a feature of the payment module.

This buyer runs bi-weekly check runs against Bank of America accounts and needs a positive pay file exported post-run in BoA's fixed-format specification. Positive pay file generation is downstream of paper check issuance: checks must be printed and numbered, and the resulting issuance file (check number, amount, date, payee) must be formatted to BoA's CashPro layout before upload. Spendesk's payment architecture does not include paper check issuance at all. Its documented payment methods are wire transfers (SEPA XML for European entities, SWIFT, and ACH/wire for USD transactions via Sutton Bank), virtual cards, and debit cards. Since the check payment run step does not exist in Spendesk, there is no check issuance record to export, and therefore no mechanism by which a Bank of America positive pay file could be generated.

For a $120M services company with subcontractors, utilities, and professional services vendors flowing through Sage Intacct, 1099 preparation requires automated vendor classification, cumulative threshold tracking, and direct IRS e-filing. Medius does not provide this capability. Medius's compliance architecture is built entirely around VAT, GST, and global e-invoicing mandates: its compliance page explicitly describes coverage as 'VAT and GST compliant in Tier 1 and Tier 2 countries' and focuses on European regulatory frameworks such as France's PDP mandate and UK Making Tax Digital, with no reference to US 1099 obligations, W-9 collection, TIN validation, or IRS FIRE/IRIS e-file integration. Third-party comparative analysis confirms that 1099 tracking and year-end filing are attributed to other AP platforms as a differentiating feature, while Medius is characterized by its ERP connector depth and European compliance coverage. The buyer's 1099 obligation would remain entirely with Sage Intacct's native module and whatever manual reconciliation work that entails, with no automation lift from Medius.

This $120M services company needs 1099 automated classification, threshold tracking, and electronic filing built into the AP workflow, so that vendor payments to subcontractors and professional services providers are monitored against IRS thresholds and filed without manual year-end reconciliation. Spendesk's fact sheet contains no mechanism for any of these requirements: there is no W-9 collection, no TIN validation, no vendor tax classification engine, no cumulative payment threshold tracker, and no IRS FIRE or e-file integration referenced in the primary or supporting tiers. Web searches of support.spendesk.com for '1099,' 'vendor tax,' and 'tax compliance' returned no results. Spendesk's own published AP automation content frames compliance entirely around European obligations (UK VAT/MTD, EU e-invoicing formats), and third-party comparative analyses consistently position the platform as designed for European-market regulatory requirements rather than US IRS reporting.

For a $120M services company processing 1,800 invoices per month across two Sage Intacct entities, 1099 compliance requires three integrated capabilities: automated vendor classification from W-9 entity type data, year-round threshold tracking aggregating payments per vendor against IRS minimums, and electronic filing with the IRS or a certified filing partner. Esker's Supplier Management module does collect W-9 documents as part of vendor onboarding verification, with one press release confirming the solution verifies 'bank details, VAT number, D-U-N-S Number, W9, insurance certificates' as part of onboarding data collection. However, no evidence exists across Esker's product pages, help documentation, or feature descriptions of a native 1099 classification engine, cumulative payment threshold tracker, or IRS e-file integration or partnership with a tax filing provider such as Sovos, Avalara, or Track1099. A third-party AP automation comparison published by Zenwork (a competitor, weight accordingly) explicitly lists Esker's gaps as 'no integrated 1099 e-filing' and 'no TIN match,' which is consistent with the absence of any mechanism in Esker's own documentation. W-9 document storage at onboarding is a prerequisite for 1099 preparation but is not the same as the automated classification, aggregation, and filing workflow the buyer requires.

For a 3-person AP team manually keying 1,800 invoices per month from email and mail today, Medius Capture delivers learning at two complementary layers. First, a network-effect foundation: SmartFlow, a proprietary CNN model, is trained on the buyer's own historical actions and enriched by 2.4 billion+ invoice field data points across Medius's global customer base, with over 393 million of those data points drawn from real-world human corrections that capture long-tail edge cases generic AI models miss. Second, a per-supplier, per-company feedback loop: learning is triggered when a user clicks 'Send to workflow' in Capture Verify, at which point the AI and ML engines ingest the corrected field values and connections the user wants the system to retain. This is not a global shared model only; the system automates repetitive corrections so that after just two corrections, it automatically applies the learned value to future invoices, and this learning is saved at the supplier level for each company, applying to both PDF and e-invoices. Beyond field-value corrections, machine learning models recognize and adapt to different invoice formats, improving over time with every document processed, and instead of relying on fixed templates, AI interprets context and patterns to identify key information, with agentic AI autonomously adjusting to new invoice formats or anomalies. The architecture is grounded in a decade of operational data: Medius's event-driven core captures all actions that humans have taken to correct errors, manage exceptions, and handle corner cases, and for the last 10 years those 'human in the loop' events have built the training dataset for their AI. This capability operates at the first stage of the pre-processing journey (invoice capture and data extraction) and feeds directly into SmartFlow's downstream GL coding and routing precision.

For a multi-location services company currently keying invoices manually from email and mail, Esker's Synergy AI operates squarely at stage 1 (capture and legitimacy) of the pre-processing journey: it ingests invoices across all channels the buyer currently uses (email, mail, EDI, fax) and applies machine learning and deep learning to extract and populate key fields into a validation form. The critical learning mechanism is explicit and documented: Esker Synergy 'improves data extraction by auto-learning from user corrections, while also utilizing historical data analysis to predict outcomes,' and the Esker blog confirms 'the AI continues to learn from user corrections and continuously improves its prediction accuracy,' noting that after only a few repetitions, non-PO invoices are automatically allocated to the correct G/L account, cost center, cost type, and tax rate. The technology stack combines deep learning for first-time recognition of new vendor layouts, machine learning for ongoing adaptation to this company's specific vendor invoice formats, OCR, RPA, and NLP under the Synergy AI umbrella. A separate 'teaching' capability is documented alongside first-time recognition and machine learning, which allows trained users to explicitly show the AI engine how to recognize specific recurring document formats, accelerating the lift curve for the buyer's 1,800-invoice monthly volume. Esker targets 80%+ touchless processing as invoices improve in recognition accuracy over time.

For a $120M services company processing 1,800 invoices per month across varied vendor formats, Spendesk offers two overlapping learning mechanisms. First, on the GL coding side, Spendesk documents a machine learning capability that predicts the expense and VAT account for each payable in real time by leveraging your historical bookkeeping data on Spendesk; this is a tenant-specific model that improves as your AP team confirms and corrects coding decisions. Second, on the extraction side, AI-powered OCR extracts supplier names, amounts, PO numbers, VAT details, and line items, and machine learning suggests GL codes based on vendor identity and transaction history, growing more accurate over time as it processes more of your invoices. A Spendesk blog article on invoice OCR also states that when your team corrects a field, the system applies that correction to similar invoices going forward, so extraction accuracy improves with use. However, this correction-propagation claim is published at blog level only; no help center documentation was returned confirming the mechanism, its tenant vs. global scope, or per-vendor template specificity. Additionally, Spendesk's invoice OCR is a feature within broader spend management rather than the platform's primary focus, and companies wanting specialized invoice processing with advanced multi-language capability or bulk PDF handling may find dedicated OCR platforms more suitable.