What is AI-powered automation?

AI-powered automation refers to software that reads business documents in the same way a person does, then takes appropriate action. This technology can handle invoices in any format, match them to purchase orders, post them to accounting systems, and flag any discrepancies. For instance, when Uber migrated its supplier invoices to this system, its engineering team reported a 70 percent reduction in the average time taken to process an invoice.

Most businesses already automate certain tasks, such as posting an entry in an accounting system when an invoice is approved or using a spreadsheet to sum up a column of numbers. However, this older form of automation follows strict rules and struggles when faced with unexpected data, like a supplier sending a phone photo of an invoice instead of a spreadsheet.

AI technology differs by taking over the reading aspect, which traditionally required human intervention. Modern AI can process scanned invoices, shipping documents, foreign language purchase orders, and customer emails to extract relevant details such as supplier names, amounts, dates, and item codes. Once these details are extracted, predefined rules determine the next steps, which may include updating systems or flagging cases for human review.

Gartner predicts that by 2026, up to 40 percent of enterprise applications will incorporate task-specific AI agents, a significant increase from less than 5 percent in the previous year. In their AI and machine learning development work, they build agents that perform tasks traditionally handled by humans and pause at points where human decision-making is still required.

AI-powered automation compares favorably to traditional rule-based automation systems. While rule-based systems can only read data in a specific format and rely heavily on human intervention for unusual cases, AI-powered automation can handle a wide variety of document formats and only requires human input in specific situations. For example, AI agents can read, match, and post invoices, taking about 25 to 30 percent less time and costing up to 25 to 30 percent less than manual processes.

Companies across various industries are quickly adopting AI automation. Uber, for example, uses AI to read and process invoices in over 25 languages, resulting in significant cost savings. C.H. Robinson, a leading freight broker, employs AI agents to handle shipment orders received via email in just 90 seconds, a significant improvement from the previous 4-hour wait times.

The banking sector has also benefited, with JPMorgan Chase implementing a contract system that processes 12,000 commercial credit agreements annually in a fraction of the time it previously took.

When starting with AI automation, it is essential to begin with processes that involve humans reading and checking documents, such as shipping paperwork or supplier purchase orders. These processes are frequent enough to demonstrate the benefits of automation but not so complex that they overwhelm the system. By focusing on these areas, businesses can effectively leverage AI to improve efficiency, reduce costs, and minimize errors.