Even in today's era, we’re sure many of you have felt that you were performing a task that, ideally, should be automated to save time and speed productivity.
We’re referring to simple activities like entering data from an Excel file into an online form, extracting data from an invoice and processing it, creating new accounts, manually depositing checks and other interactions with application user interfaces.
Essentially, any task in which a human interacts with an information system is a good candidate for a bot (software robot) to automate. But what role does artificial intelligence (AI) play in this process? Which AI capabilities help drive the business value of robotic process automation (RPA)?
A sample use case
First, let’s see how easy it is to create an RPA digital assistant. In this video, we will author a bot that can extract information from an invoice and then use that information to update a database and send email notifications.
Now, let’s explore how AI technologies such as natural language processing (NLP) and optical character recognition (OCR) make robotic process automation possible.
Artificial intelligence in chatbots (NLP)
One of the most common types of RPA bots today are chatbots. Whether it’s paying a bill by phone or interacting with an automated customer service agent, you probably have some experience with chatbots. This technology still has lots of room for improvement. However, the nature of chatbots is that the more we train our models, the more effective their engagement with customers will be.
A big part of AI is natural language processing (NLP), which also plays a key role in the operation of chatbots. With IBM® Robotic Process Automation, users can train the AI models that are the backbone of the chatbot by uploading a knowledge base, which consists of a basic set of questions and answers. IBM RPA uses that knowledge base as a training tool to help the chatbots interact with users. Using AI with NLP, chatbots can then respond and trigger other actions (e.g., opening a ticking, making a travel reservation) in a more meaningful and natural manner to the questions it receives from humans.
Artificial intelligence in computer vision
Computer vision has long been a compelling topic of interest in the computer science community. It has a large set of applications, including vehicles that are capable of autonomous driving. A variety of machine learning algorithms are at play in computer vision, such as neural networks, which are a backbone for deep learning.
However, one very practical application of computer vision is the ability to extract data from scanned text documents, non-fillable PDF files and various other documents. This is made possible by optical character recognition (OCR). The bot in the demo video above uses OCR in IBM RPA to extract data from the invoices. OCR is an effective tool for automating and accelerating processes by reducing the need to complete them manually.
Bringing OCR and NLP together
When these two powerful capabilities come together, they offer tremendous business value for a variety of use cases. One example is document classification. After scanning a document using OCR, the bot can then analyze the set of fields and text via NLP and classify which type of document it is (e.g., a purchase order or an invoice). It can then route the document for processing to an appropriate backend system. If the bot cannot classify the document, it will initiate human intervention via a Slack message, email or other communication.
Another related use case is the processing of legal documents. A bot can leverage the AI capabilities of OCR and NLP to analyze various clauses and ensure that they are compliant with current rules and regulations. The bot can also check the document for completeness, such as making certain that signatures are present at the right location and validating those signatures based on previously scanned images.
Get started with IBM Robotic Process Automation
As you can see, there are many use cases and possibilities for using bots with robotic process automation to accelerate everyday tasks. The potential value for businesses that take advantage of these capabilities is huge, and artificial intelligence is the primary enabler for many of the capabilities provided by RPA.