Manual invoice reconciliation is a hidden bottleneck for many teams. Employees spend hours manually cross-checking invoices against purchase orders, to ensure items, prices, dates and quantities match. The result? Delays, errors, and frustrated teams.

‍

This is the most common outcome for just one case of reconciliation – i.e. with purchase orders, but imagine the effort and the results when you have to match invoices with delivery notes and warehouse logs. Things get unnecessarily complicateIn this guide, we’ll walk you through how to automate the reconciliation process using Nordoon AI Agents, that turni the time-consuming task of matching invoices with purchase orders into a seamless workflow.

‍

Step 1: Upload your purchase order data

Start by uploading a file that contains the items from your purchase orders. This dataset will serve as your lookup table, i.e. the reference point Nordoon AI Agents will use to verify invoice data.

‍

‍

Step 2: Customize your invoice template

In the Nordoon app, choose a pre-built template for invoice processing. You'll customize this template to fit your workflow:

• Step 1 (customer info): Define basic invoice and customer metadata.

• Step 2 (line-item checks): Create rules to match invoice data against your PO records.

‍

To improve accuracy, connect two lookup match keys:

• PO number (from the invoice and PO table)

• SKU code (from both sources)

‍

‍

Now, define rules for critical fields:

• Quantity: Match quantities from the invoice to those in the PO. If mismatched, have the AI agent flag it.

• Item price: Set a rule to compare pricing data and flag discrepancies automatically.

‍

Set both rules to "require" so the AI agent can flag any inconsistencies that must be reviewed.

‍

‍

‍

Step 3: Add a VLOOKUP step

Next, create a VLOOKUP step to retrieve item details directly from your PO table. This helps the AI agent pull additional context about the products being verified, enriching the comparison.

‍

‍

Step 4: Configure an AI summary

Now for the human-friendly part: an AI-generated summary.

‍

‍

Once your template is saved, configure a context prompt that tells the AI agent what kind of summary you expect – i.e. a concise explanation of mismatches or confirmation that all fields are aligned.

‍

Step 5: Automate supplier communication

Finally, automate the response back to your supplier.

‍
To reduce communication overhead, you have the possibility to ask the AI agent to generate an email summarizing the outcome of the reconciliation. Under the guidelines section, you can prompt the Agent to highlight any inconsistencies clearly.

‍

‍

Once that’s set, link the summary and email steps back to your invoice step (step 2), so the workflow remains logically connected.

‍

‍

Final step: Test the automation

Upload a sample invoice and run your template.
In the Result, you should see:

• Flagged mismatches in quantity or price

‍

‍

• A clear AI-generated summary

• A ready-to-send supplier email, which can be automated further

‍

‍

And the best part? You only need to build this template once.

‍

With Nordoon AI Agents, invoice reconciliation becomes a one-click process. No more manual checks. No more cross-referencing hundreds of documents. Just clean, reliable, automated validation powered by AI.

‍

Try it yourself and see how much time (and sanity) you save.