AMD transforms accounts payable process with artificial intelligence

Automated workflow

We configured the workflow for prompt routing of invoices and automatic generation of email alerts to person(s) responsible for approving invoices. When an invoice is received, the AMD SAP FICO enterprise system triggers an email to the approver, based on pre-defined date range and approval limits. The response (approval / rejection) and the email id of the respondent is automatically routed to an Infosys NIA shared mailbox. An ‘intelligent’ robot in the Infosys NIA system extracts and analyzes data in the invoice mail box at intervals of five minutes. It ‘reads’ messages and captures relevant information such as the email id, workflow details, invoice number, action, and comments.

Infosys NIA validates the user id and workflow-user mapping. Subsequently, it updates the invoice status as approved or rejected in the SAP system. Infosys NIA then receives a success / failure reply from the SAP Remote Function Call (RFC) interface system and transmits a confirmation email of the action (invoice approved / rejected) to the approver. Finally, Infosys NIA moves the approval mail from the inbox to the archive folder in the NIA mail box.

On-the-go access

The AMD team is empowered to approve invoices using mobile devices. Business users need not log into the SAP portal or connect to the enterprise VPN network for accessing and approving invoices. A soft copy of the invoice is made available in the notification email, which ensures instant access to relevant data for informed decision making. The system also sends a periodical reminder email to approvers for each pending invoice.

The Infosys AI solution boosts productivity by aggregating data and eliminating the need to navigate between screens to extract information for taking appropriate action. In addition, our solution offers a consistent experience to AMD’s AP team and executives, across mobile devices.

Agile development

Infosys adopted the agile methodology to maximize value of the AI solution. Our approach enabled both our IT teams and AMD’s stakeholders to refine business requirements, identify process / workflow improvements, and assimilate feedback from users.

During the pilot, the solution was rolled out to a small group from the AP team who approved invoices. It helped the Infosys team identify issues in real time and fix them by either improving the functionality or adding new features. We undertook concurrent development and testing until all business requirements were addressed and the solution reached a stable state. Consequently, it was rolled out to global users.