AI Enablement For Finance Teams , AI Enablement For Sales Teams , Google Sheets Automation , How-To , Operationalize AI For Business , Zenphi Use Cases By Industry Or Role
Step-by-step tutorial on the most budget-friendly accounts payable workflow. How to automate invoice processing, handle accounts payable automation, unstructured data extraction and routing using Ai and Gmail.
If your goal is to build a resilient profitable business, streamlining operations is crucial for efficiency and growth. One area ripe for optimization is intelligent invoice processing and accounts payable tasks. By leveraging the power of automation, organizations can significantly enhance their accounts payable workflow, saving time and resources. AI invoice processing tools enable seamless automation of invoice data extraction, even from unstructured sources, leading to more accurate and efficient processes. In this blog post, we’ll delve into the benefits and strategies of automating invoice processing and accounts payable processes, exploring the transformative potential of accounts payable automation.
Table of Contents
Who would benefit from accounts payable automation
Almost any business dealing with a steady stream of invoices, particularly those subject to high volume fluctuations, stands to gain from automating their invoice processing. Whether you’re in wholesale, retail, logistics, catering, or reselling, prompt processing of supplier invoices, timely payments, and accurate markup billing for buyers are essential for maintaining positive cash flow. Take, for example, Zephi’s success story, where a client managed to process six times more invoices with our assistance without expanding their workforce. This underscores the significant efficiency boost achievable through automation.
So, if the results sound appealing to you, let’s dive in the automation tutorial.
Step 1: Create An AI-Model In Zenphi
Log into your Zenphi account and navigate to the AI Builder tab. Select the Invoice Parser for this specific scenario.
The model will prompt you to upload a test invoice to train it.
In other words, you need to map the fields from the invoice to the data categories you want to extract.
Zenphi’s AI model builder offers a wide range of standard variables relevant for invoices: due date, supplier, total amount, and more. Choose the ones you want to map or explore additional fields from other document types.
Test the model with another invoice to ensure all fields are mapped correctly.
Step 2: Create a Process Trigger — Incoming Email with an Invoice
Return to the main dashboard and begin creating a flow. While you can choose other triggers to initiate your automated invoice processing, starting with an incoming email containing an invoice from a supplier seems reasonable.
To configure your trigger, set up conditions to ensure the process monitors only the accounts payable mailbox and deals with emails containing at least one attachment.
Step 3: Query Your Collection
As with any process automation, you need to query your data to begin using it in subsequent steps.
Step 4: Run Your AI Invoice Processing Model on the Collection
In this step, run the model you created earlier. The results should include a set of variables that your AI invoice processing model extracts from the invoice, such as the supplier’s name, due date, and total amount.
Zenphi’s AI model provides a confidence level, allowing you to structure the flow so that only invoices with a confidence level above 85%, for example, are sent to the next stage automatically.
For invoices where the model is less certain about extracted unstructured data, you can set up a different flow, such as sending an email back to the supplier to clarify details or confirm the extracted data.
Assuming the model extracted the data correctly, proceed to route it further.
Step 5: Check If Supplier Exists
Some suppliers may already be listed in your CRM, while others may be new. Include a step in your automated accounts payable workflow to check for supplier existence.
You can easily build this flow in Zenphi using a Lookup function to search in the specified database.
Structure it as an If condition: lookup the supplier name extracted by the AI model; if it exists, proceed with the data; if it doesn’t, we recommend adding a step to automatically request a manager to register a new supplier in your system.
Step 6: Approve The Invoice With a Cost Center
Assuming the supplier exists, include a step in your accounts payable automation process to verify the invoice with the relevant cost center. Again, use a lookup function to check the cost center associated with the supplier and their email. In this example, Google Sheets is used as a database, but Zenphi can also lookup data in HubSpot, Salesforce, Monday.com, and ClickUp. Send an automated email to the relevant cost center asking them to approve the invoice using Asign Task function.
In this case, the relevant cost center will receive a form like this that will allow them to take action easily.
Step 7: If the Invoice is Approved, Send to the Approved Invoices Database
Add the next step in Zenphi. If the invoice is approved, you want a new row added to your Approved Invoices database. In this case, Google Sheets is used as the data source, but Zenphi can also add subitems to multiple CRM systems. Structure this step so that your database contains all data extracted by the AI model from the invoice.
Additionally, notify your accounts payable that the invoice is approved and should be paid. This can be done by adding a step in Zenphi to send an email via Gmail or using labeling in Gmail to send a new email labeled “Approved Invoices” to your accounts payable.
Finally, using Generative AI, you can automate the creation of a markup invoice for a buyer based on the extracted data and commission data stored in your CRM, and email this invoice automatically.
However, this will be covered in one of your next tutorials. For now, watch how our colleague, Steve, goes through the process step by step.
Talk to us
Using a more refined procedure to process AP and invoices? Let us help you to build the automated flow to support it!
