Intelligent Process Automation (IPA) is not one technology, but a collection of emerging technologies that come together to automate and improve operational efficiency of a business process. It is a combination of fundamental process redesign combined with leading technologies in automation and Artificial Intelligence to take the “robot out of human” and enable knowledge workers to focus on more high-value activities. IPA is a methodology to help enterprises radically improve customer journeys, simplify interactions, and speed up processes.
IPA shares its overarching goal with RPA, IDP and other automation technologies to automate mundane repetitive tasks in a business process, but at a different level. While RPA, IDP and others focus on automating a task at a time, IPA is focused on automating several tasks in a business process workflow to achieve higher level of automation and efficiency. If done right, IPA can radically enhance operational efficiency, increase worker satisfaction, reduce risks, and improve customer journeys.
IPA encompasses a set of core technologies:
-
Robotic Process Automation (RPA)
RPA is a software tool that can be configured to extract data from structured data sources, manipulate, trigger responses, and communicate with other digital systems. RPA technology is inherently rule based and can automate mundane activities of interaction with several digital systems as part of a business process. It is non-invasive in nature and sits on top of the existing systems. The “bots” effectively mimic a human worker with their own user id and password.
-
Intelligent Document Processing (IDP)
IDP (along with RPA) falls under the same umbrella of “Information Extraction”. However, IDP is targeted towards extracting information from unstructured data sources like, document, emails, images, etc. IDP combines the power of AI Machine Learning technologies to process unstructured content and extract information. With recent advancements in Natural Language Processing (NLP) and Computer vision technology (OCR), IDP is coming into mainstream adoption and automating mundane work of data entry tasks into back office corporate systems. Read RDP vs IDP for more detailed comparison of the two technologies.
-
Smart Workflow
These are process-management software tools that orchestrate various steps in a business process. The tool keeps track of each step in the workflow, manages hand-offs between different groups (including bots and humans), and provide statistical data for ongoing fine tuning of the process.
-
Natural Language Generation (NLG)
NLG is cutting edge technology from AI Machine learning space. NLG engines can interpret and translate observations from data into natural language text that can then be consumed by humans. NLG is used extensively by broadcasters to draft stories about games and markets in real time. Financial services firms use NLG engines to generate weekly management reports by feeding structured data to them from internal and external data sources.
-
Cognitive Agents
These are virtual agents that combine machine learning technologies (specifically NLP and statistical) to augment human workforce. Cognitive agents can communicate, learn from the data in an on-going manner, conduct sentiment analysis and execute tasks on their own in certain scenarios.
-
Machine learning / Analytics
ML and Advanced analytics are set of algorithms and mathematical tools that power many of the intelligent behaviors of above-mentioned technologies. In that sense, ML and advanced analytics is a horizontal enabler that can be implemented across the entire business spectrum to address specific challenges and issues, like fraud detection. The ability to consume large amounts of data and generate insights or make predictions makes it a powerful tool to have in the bag.
Case in point: Claims Processing
At a typical insurance company, human claims processor refers to more than 10 disparate systems to process a claim. With IPA:
- NLP can scan the request and kick start the claims process (IDP)
- Robots can replace manual clicks to collect data from disparate systems for analysis (RPA)
- Machine Learning models can assess acceptance given the historical data (Machine Learning)
- A chat bot can clarify customer queries (Cognitive Agent)
- Summary report can be produced for a human to review (NLG)
Conclusion
The full range of benefits come by implementing the complete suite but need to be carefully orchestrated for success. The entire initiative needs to have a well thought out strategy, coordinated across multiple functional departments, redesign processes as needed, and should be very well supported with a deep knowhow of individual technologies. Organizations should do a thorough evaluation of their business objectives, current market conditions, time frame for implementation, emerging technologies, and risk tolerance before embarking on an end-to-end IPA initiative. Significant benefits can be realized by implementing individual technologies to remove bottlenecks in business processes which may be quick to implement and can be built upon incrementally. Take your pick.