Organizations are increasingly automating physical and knowledge work for efficiency and productivity gains. To benefit from automation, organizations must first determine what kind of automation solutions can meet their requirements and goals.
Robotic Process Automation vs. Business Process Automation
If you can identify tasks that software can take over from human resources and perform more accurately, consider robotic process automation (RPA) solutions. On the other hand, if you want to automate processes across the organization with the goal to optimize them, assess business process automation (BPA) solutions.
Unlike RPA, which involves automating specific tasks within an existing business process, BPA automates complex processes. What is common to both is that they replace manual effort.
So, if your goal is to optimize existing processes or a specific, repetitive task, consider an RPA application or software. However, if you’re interested in end-to-end automation of a process or workflow, or want a BPA fix that must be designed from the ground-up, choose an artificial intelligence solution.
BPA uses artificial intelligence while RPA doesn’t
The term RPA is misleading as it is not a robot but a software that autonomously executes a task based on predefined business rules. Think of an application that retrieves invoice emails, downloads the attachments, and creates bills. These tasks require no intelligence; they replicate repetitive human actions – retrieving, downloading, and copy-pasting.
Artificial intelligence involves thinking and learning. AI reads the invoices and extracts relevant information from the semi-structured data. It correctly deciphers the information to be extracted from a variety of invoice templates and formats.
If you’re an insurance company, BPA can automate your entire claims processing, while RPA can be used for one or more tasks within that process, such as moving records from one database to another.
In essence, RPA is process-driven while AI is data-driven.
BPA is big-picture oriented. It employs automation where needed while also allowing integration with external technologies and increasing the efficiency of humans involved in the process. In automating end-to-end processes, BPA:
- Identifies all administrative tasks around the process
- Outlines people’s workflows
- Pinpoints areas in which those workflows can be optimized
Capabilities of AI
The three key AI capabilities are:
Machine learning is a capability of AI that allows machines to learn from examples and act without being explicitly programmed. Computers can process large volumes of data and connect the dots across thousands of variables, something even the smartest humans cannot achieve with a high degree of accuracy. The ability to draw insights from millions of data points quickly reduces the duration of tasks. For instance, machine learning assists attorneys with legal research, saving them time for more in-depth analysis and case development.
Natural language processing
NLP is the capability that powers human-machine interactions. It gives machines the ability to read, understand, and derive meaning from human languages. From the perspective of business process management, NLP can glean business intelligence from terabytes of data. Other applications include extracting information such as names and account numbers from forms and text resources, grouping forum discussions by topic, and linking entities.
The healthcare industry may use NLP to analyze text data in clinical forms and patient notes, while law enforcement can leverage data from criminal records, social media posts, and anonymous phone calls to speed up investigation.
IBM Watson is a well-known example of NLP at work. The question-answering system can draw context from statements by examining words, tone of voice, placement of terms and word choice, making sense of business queries to make better-informed decisions.
AI solutions possess interactive capabilities that allow natural, seamless human-machine conversations. In other words, there is no need to define specific requirements and scenarios to enable interactions. For instance, an AI solution can engage a web visitor with questions such as ‘What color shoes do you want to buy?’ or ‘What occasion do you need to wear the shoes to?’ and relevant, purchase-related queries that help potential customers on their purchase path.
Conversational marketing is one way in which this AI capability is being applied. Chatbots can interact with hundreds of website visitors simultaneously. A conversational marketing platform may also integrate with the enterprise user’s analytics platforms, CRM system, and IT/telecommunications systems.
AI can improve business process management
AI can support BPM initiatives, by automating routine tasks, analyzing large amounts of data, and enhancing user interfaces. Data analysis, machine learning, and predictive analytics can automate several of the basic decision-making processes in an organization. According to McKinsey, AI can automate up to 45 percent or more of a particular job, freeing up workers for more mission-critical or high-value work that technology cannot easily accomplish.
Here are three use cases:
- An AI tool can monitor regular business processes to note where issues may potentially crop up. It may include an ‘assign’ feature where the system decides who needs to act on a request, based on current availability and past performance. The tool can also analyze business processes to make recommendations on process optimization.
- An AI tool may optimize workflows to avoid spending more time than necessary on tasks that don’t require this extra investment. It can analyze the time workers need to perform specific tasks. By applying machine learning, the tool can identify data, forms, and situations that lead to a higher-than-necessary time investment. For instance, it can make suggestions such as ‘Do you want to auto-approve similar requests?’.
- AI has many applications in improving customer support. One use case is to automate voicemail follow-ups, ensuring that customers who leave a voicemail are engaged quickly and receive appropriate responses to their queries. Such an intelligent tool could use speech to text conversion to create a transcript of the voicemail, and then take the text and add it to the service ticket, offering customer support agents a clear idea about what the complaint entails for quicker resolution.
Leveraging AI for BPM
Understand the problem you want AI to solve
AI for BPM should deliver real value to the organization and its employees. It requires an in-depth analysis of business problems so as to offer meaningful solutions tied to definite benefits. Organizations can form a committee to identify areas where BPM can improve efficiency, accuracy and/or productivity, and then explore the kind of solution(s) that can achieve those gains. Only after a thorough examination of these factors can the AI solution be designed from the ground-up.
Avoid taking on too much too soon. Identify low-hanging fruit – processes that may benefit most from automation. As AI applications need clean data, you could start with a process where clean data sets are readily available. An incremental use of AI will still give you valuable feedback that you can use to expand as needed. If clean data is unavailable, assign the committee to sort out inconsistencies in the data sets of the targeted processes, so that you can start with accurate and rich data.
Encourage the adoption of AI systems
Enterprise AI adoption has increased 250 percent over the past four years, although adoption remains uneven across sectors. As AI adoption for BPM is an organization-wide endeavor, it is imperative that people feel prepared for the changes it will bring to their routine, and are aware of its benefits to them and the organization. Along with integrating technology into work processes, see if you have a culture that is primed to increase human-machine interactions.
Address internal capability gaps
Before pursuing a full-scale AI implementation, organizations need to review their internal capabilities. Are processes adequately evolved? If not, you should first focus on addressing gaps by involving teams and leveraging existing projects.
What to keep in mind when embarking on business process automation
- Make sure automation enhances the ‘meat’ of the targeted process. Automating workflows requires understanding them deeply on a step-by-step basis. Define the entire process in specific, precise steps to ensure that the right conditions trigger actions in the correct sequence.
- Ensure that you choose the right IT consultant who can offer guidance on selecting solutions that are adequate, suitable, and do not pose integration challenges. Integrating various tools and systems is a challenge of implementing business process automation. Only if the chosen systems ‘talk’ to each other can you realize the full benefits of BPA.
- Understand the regulations you might need to meet concerning data protection and security. Also, review the impact of company policy to ensure that there is no misuse or compliance issue.
- Follow-up on the productivity, efficiency and/or cost benefits of implementing AI. Establish pre-automation benchmarks to determine the extent of gains post-automation. Also, seek feedback from employees on their experience using the AI systems and its impact on their work.
Finally, as AI often creates a fear of job loss, you should effectively convey that automating business processes can open up opportunities for employees. It is up to the organization to remove ambiguity around AI adoption and sell the benefits of business process automation to the workforce.