Automation Academy

The Role of Manual Tasks in Intelligent Automation

5 min read
Manual Tasks in Intelligent Automation

As the popularity of the terms “artificial intelligence” and “intelligent automation” rises, so does fear of the changes they can bring into our future. One of the most common concerns people have is how automation is going to affect our jobs. The fear of unemployment, especially in a time of frequent economic crises, is understandable. However, there is also an opinion that robots will complement humans in their work rather than replace them.

Combining the power of people and software robots to automate a manual process can lead to great, often even unexpected, results and positive changes in the company. WorkFusion’s Intelligent Automation Cloud provides capabilities for including special manual tasks in your automated workflows (also known as “human in the loop”) that allow employees to participate in the automation process — thus bringing more benefits of automating manual processes to the company.

Benefits of person-bot cooperation

First, let’s understand why combining the work of people and bots is so beneficial.

The main reason is that people and bots are good and bad at different things. Thus, when working together in the automated process, they compensate for each other’s shortcomings. Bots are great at working with large amounts of data and repeating the same tasks without getting tired and thus making mistakes, while people add qualities such as decision-making, improvising, and innovation to the process.

Zero to OneA great example of such work was described by Peter Thiel in his book, Zero to One: Notes on Startups, or How to Build the Future. In mid-2000, PayPal was processing thousands of transactions per minute and losing $10 million every month due to credit card fraud, as the sheer volume of the transactions made it impossible to review them all via manual processing. So, Max Levchin’s team of elite mathematicians devised a sophisticated algorithm that detected fraudulent transactions. It didn’t work well, though: Fraudsters adapted and changed their methods quickly and fooled the system easily. However, it turned out that fooling people wasn’t as easy. So, the ultimate solution to this problem was creating a hybrid fraud detection and prevention system, called “Igor.” The algorithm quickly detected and flagged suspicious transactions, and a team of human analysts reviewed them. This system eventually helped PayPal to turn their losses into profit in 2002 and was later used by the FBI to detect financial crime. It shows how people and robots can cooperate and complement each other to provide better benefits of process automation.

Now, let’s see how the human-in-the-loop approach is utilized in Intelligent Automation Cloud through manual tasks.

Using manual tasks in WorkFusion

There are several tasks that human-in-the-loop solves in automating operations, depending on the automated manual process and the technology used. We briefly covered how manual tasks can be used in our RPA tool in one of our previous posts. Now, let’s see some of these and other cases in more detail.

Exception handling

Exception handling is one of the main purposes of using manual tasks in an automated business process. In cases when a bot cannot successfully perform some actions, a human can enter the workflow and either perform the task the bot failed in, or correct the circumstances that prevented the bot from finishing. For example: correcting an error in the data the bot draws from needs. Let’s see two examples of how a manual task can be used in intelligent process automation for exception handling.

Example 1: The bot reads some currency codes from the file, gets the USD equivalents using an online currency converter, saves the results in an Excel report, and sends it via email. If there was a problem with the converter website or a wrong currency code was provided, the bot would not be able to get the correct results for some currencies online. In this case, the bot would transfer these currencies to a person via a manual task, wait till the person finds and provides the correct results, and resume the work.
Example 2: An RPA bot extracts information from hundreds of invoices using OCR and creates a report in Excel. Some of the invoices are non-standard, so it cannot extract the required information and sends such invoices to a person in a manual task, waits till the person tags the information in the documents, and then uses it in the report.

In these cases, human-in-the-loop increases the benefits of automating a manual process by combining the best qualities in bots and people: a bot’s ability to process a large amount of data, and a person’s critical thinking.

Making decisions

This is another important purpose of a manual task. In some cases, you can automate the manual process from beginning to end and provide several courses of action for the bot to perform depending on certain conditions. However, the rule of following a particular course of action cannot be defined easily and requires a non-standard judgment. In this case, you can involve a human in the decision-making process through a manual task.

Example 3: The bot reads an Excel report containing information about the company’s invoices, updates the information in various databases and sends reminders or thank you emails to the customer depending on the status of the invoice (paid/unpaid). Sometimes, the status column will contain some notes from the account manager, and additional evaluation of future actions regarding this customer is required. In this case, the bot can send the information about the customer and the invoice to a person via a manual task, and the person will evaluate all required information and select the course of actions: send a reminder, ignore, notify the financial department, etc. Depending on the person’s choice, the bot will perform different actions.

Training cognitive bots

Another important function of a manual task in Intelligent Automation Cloud is training the cognitive bots. Machine Learning is an integral part of our Enterprise Edition, and manual tasks provide the user-friendly interface for people to both train the bots to process documents that contain unstructured data and to extract the information required in the automated processes and review the results. Using this approach allows users to implement cognitive automation in the company without the need to involve a team of data scientists.

These are just the most common applications of manual tasks in Intelligent Automation. To learn more about human-in-the-loop and other components, sign up for free community courses in Automation Academy today!

Start your Automation journey