Rossum Integration (4): Changepond Automates AP Data Capture with Rossum

Our Rossum integration success story for this week comes from Changepond Technologies. Established in 2000, this “Native Digital Integrator” has been helping companies manage the convergence of enterprise software and data solutions. You’re about to learn how Changepond and one of its key business partners, a provider of back-office accounting services, are working with Rossum to transform the accounts payable (AP) process for well over 1,500 businesses.

Changepond

Introduction 

The AP process is immensely important to your business because it involves nearly all your company’s payments outside of payroll. Regardless of your organization’s size, AP’s main responsibility is to manage invoice processing, approval, and payment. During the processing stage, AP must check the legitimacy of the invoices as well as the accuracy of the data captured from them. 

Invoice formats vary among vendors; therefore, the data capture process could prove to be challenging and time-consuming for your AP team. This can have an impact on accuracy, which must be high to ensure error-free processing.

Changepond and Partner 

Changepond provides its partner with automation solutions for AP and other back-office operations including, but not limited to: 

  • Accounts receivable
  • Expense management
  • Debit and credit card statement processing
  • Payroll
  • Customer onboarding 

AP is the partner’s key service, which it provides to its 500+ clients, as well as their clients’ 1000+ vendors. The partner handles the AP process using different invoice management and accounting software platforms that vary from client to client. Functioning as a back-office team, Changepond’s partner processes 15,000 invoices every month, extracting data from each invoice then entering that data into the accounting software of its clients and their vendors. Previously, Changepond’s partner was handling these operations manually. 

The Accounts Payable Process 

Prior to automation, the AP process consisted of the following steps: 

1. Changepond’s partner received invoices from its clients and each client’s vendors as email attachments or email content, or through invoice management software like SAP Concur or AvidXchange, file-sharing solutions like ShareFile or Dropbox

2. The partner downloaded invoices then extracted key data, including:

  • Invoice date 
  • Invoice number 
  • Vendor name 
  • Vendor address 
  • Payment terms 
  • Line items 
  • Tax 
  • Subtotal amount 
  • Total amount

3. The partner’s accounting team then entered the captured data into their clients’ respective third-party accounting systems, e.g. QuickBooks, Bill.com, NetSuite. 

  • The partner created digital invoices for vendors that already existed within the clients’ accounting systems, then assigned relevant general ledger and class codes, delegated responsibility for approval, and submitted the corresponding records. 
  • For new vendors, the partner added vendor details to the records of their respective clients, then created the corresponding records. 

4. The partner uploaded the invoices as attachments to each record it just created. 

5. The partner then uploaded the invoices into a common ShareFile location. 

6. The partner clarified or corrected data manually with clients and their vendors, then made the corresponding updates in their system. 

Changepond – Rossum Integration 

As a partner, Changepond provides an automation solution for the AP process, using UiPath where a robot, or “digital agent”, handles most of the manual tasks. The robot is integrated with Rossum to extract and export invoice data efficiently and accurately, without manual intervention. 

Dedicated AI Engine Creation 

Rossum created a dedicated artificial intelligence (AI) engine for data capture, adding data fields to the platform’s default fields to ensure the capture of all required information from invoices. 

Additional data field for the dedicated AI engine.

Training Rossum

As a collaborative effort with their partner, Changepond used Rossum to annotate over 3,000 invoices from different vendors and submit them for approval. Changepond also trained Rossum’s AI as it annotated these invoices, using the platform’s intuitive user interface (UI) to validate and correct data capture results. This initial training period helped improve Rossum’s accuracy during future automated data extraction operations. 

Changepond annotates invoices which trained Rossum's AI.

UiPath and Rossum Integration 

Using UiPath, Changepond designed the robot to connect with Rossum using the latter’s API – the robot made API calls to establish a connection with Rossum and retrieve invoice data. Changepond used the following functionality APIs in the process:

How Changepond's robot connected with Rossum to process invoices.

In UiPath, Changepond created a separate workflow to handle all Rossum integration processes. 

Invoking the Rossum Workflow

Invoking the Rossum Workflow.

Four main process workflows are involved in Rossum integration, namely, 

  • Importing invoices into Rossum 
  • Verification and validation of the imported invoices 
  • Export extracted information from Rossum in CSV format 
  • Preparing output based on the captured data received from Rossum 
Steps in the process workflow.

The API URL to call Rossum functionalities from UiPath is https://example.app.rossum.ai/api/v1 (Read about API Base url in documentation)

  • Login – https://example.app.rossum.ai/api/v1/auth/login
  • Get Queue ID – https://example.app.rossum.ai/api/v1/queues
  • Import files – https://example.app.rossum.ai/api/v1/queues/<Queue_Id>/upload
  • Start review – https://example.app.rossum.ai/api/v1/annotations/<Annotation_Id>/start
  • Export results – https://example.app.rossum.ai/api/v1/queues/<Queue_Id>/export?format=csv&status=exported
  • Delete files – https://example.app.rossum.ai/api/v1/annotations/<Annotation_Id>/delete

API calls from UiPath are made using HTTPRequest. The robot receives responses from Rossum in JSON format for use in other business applications.

Robot compares, posts or reports data.

The robot compares the data captured by Rossum with the data contained in the corresponding client’s consolidated master invoice sheet. Upon the successful completion of this data validation step, Changepond’s UiPath robot posts the data into invoice management software and accounting systems. If there are any discrepancies in the comparison, the robot reports them to the AP team, who can correct and validate data in the Rossum UI manually.

Conclusion 

Rossum’s invoice data extraction solution provides quick, easy, and accurate invoice data capture for Changepond and its partner. Following validation and training during the initial annotation process, the Changepond robot, when integrated with UiPath, fully processes 60% to 75% of all invoices that go through the system automatically, without manual intervention.

Ready to get started?

Make a quantum leap in your document processing approach. Boost accuracy and effectiveness with an AI-powered data capture solution for all documents.