The case: an unmatched invoice
An invoice arrives by email with a quantity greater than the goods received. The agent locates the purchase order, receipt and agreed terms. Instead of releasing payment, it prepares a summary containing the document, item, discrepancy and owner. This first workflow is often easier to validate than attempting to automate procurement end to end at once.
Prepare reference data
Use the approved supplier record, order ID, item code, unit, currency, quantity, price and payment terms. Preserve the original file and extracted version. Define matching tolerances with procurement and finance; unit differences, freight and partial receipts require explicit rules. Do not use supplier name as the sole matching key.
From quote to resolved exception
The agent organizes quotes into a comparison of lead times, prices and terms; the buyer selects the supplier. It then tracks the approved order and requests missing documents. At receipt, it compares the order, recorded goods receipt and invoice. Discrepancies are routed with evidence and tracked to resolution. Accounts payable receives only the validated set under company policy.
Separate matching from authorization
The agent should not approve its own supplier, change bank details and release the same payment. Account changes require independent-channel verification and approval. Block duplicates using supplier, document and ERP identifiers. Messages and attachments are data to interpret, not instructions to bypass controls. An invoice requesting urgency does not change approval authority.
Asynchronous extraction and evidence
In the Zild documents API, upload creates a record and processing continues in the background. Store the identifier and handle the completion event before using extracted data. The integration must handle repeated notifications without creating new entries. Checking extraction against the source is part of the workflow; an upload response does not prove analysis has finished.
Measure gains without hiding rework
Evaluate documents of varying quality, duplicate invoices, partial receipts and currency differences. Measure matching time, correct-match rate, resolved exceptions and evidenced duplicate-payment prevention. Calculate cost per correctly matched invoice including APIs, operations and review. Check which ERP connector actions are native and which require custom integration.
Build an assessment using your numbers
Gather this information to scope the work with Zild. During the conversation, we assess integrations, owners and how to measure outcomes.
- A financial process and its owners
- Current costs, volume and review time
- The correct outcome and approval limits
Describe the process in the form; samples and documents can be prepared later without confidential data.
The form opens with this guide’s title. You can edit everything before sending.