Robotic Process Automation (RPA) as a concept has its beginning in early 2000s. But eventually became popular among large organizations only in around 2015. The last two years have made it a real ‘hot’ term in technology innovation, so much so that two of the top three companies in this space have become unicorns (with valuation of 1 billion $ +).
What is RPA
In simple terms, RPA refers to the type of automation which does not require resource-intensive or code-intensive integration between multiple systems. It mimics the actions of humans on different systems and enables building of virtual robots to perform the activities themselves. Now, this does not explain everything that RPA software does.
Many RPA vendors collaborate with other technologies to provide an enriched software experience. To understand the concepts, benefits and challenges of RPA better, it would be good to quickly get some basics around the following terms:
Structured data vs Unstructured Data
Data than can be searched and queried easily are termed as structured. Most common software that house structured data are relational databases such as Oracle DB and MS SQL, which have pre-defined models or schema to differentiate types of data. There are some semi-structured data holders such as XML, JSON, etc., which are not structured in the same way as databases but can help convert to structured data. Both structured and semi-structured data together constitute less than 20% of all enterprise data. So, one can imagine the potential of technology if what we have now automated is with this little data.
The remaining > 80% of data is unstructured – in other words, the different components of the data cannot be easily searched, referenced or queried. One can now realise that the rest of data types such as emails, images, videos, word processors, spreadsheets, social media, websites, mobile data, etc, all come under the unstructured category. Though some liberalists and technology vendors debate on some of these such as emails, excel, CSV, etc. as falling under semi-structured category, the purists still are not convinced.
Most enterprise technologies work with structured data. And till recently most technologies have their own native functionality for managing such data. So, any movement or operation of data between systems was a process gap and needed human or manual intervention to bridge the gap.
Integration vs Screen Scraping
The ideal way of working with processes that houses data in different software is via ApplicationProgramming Interface (API) integration. This approach enables end-to-end process management, and the outputs do not get impacted in case of any changes to the user interfaces of the systems. The data accessed from the systems mayor not be transferred into the databases of the BPM technology.
Where API integration is difficult, either due to technical reasons or costs, an alternative method is to perform surface integration from the user interfaces of various systems. This allows a logical interpretation of the different operations that are performed by users on the screens of such systems. This approach is referred to crudely as Screen Scraping. RPA has its origins in Screen Scraping technology. The latest RPA technologies have evolved from screen scraping to more advanced logic and additional features to record operations, having built-in commands for common operations, control towers, analytics and orchestration capabilities.
Attended Automation vs Unattended Automation
Automation of activities or processes without human intervention is referred to as unattended automation. Most rule-based, less complex and highly repetitive tasks can be automated in unattended mode. Back-office operational processes are good candidates, and RPA is best suited for unattended automation. Attended automation can be used when human judgement is necessary but needs enhanced support with experiential and logic-based automation. For example, in sales and contact centres where information related to the customer queries that are residing in different systems are automatically retrieved to aid the human workforce in serving the customer faster and better. A high level of observation and trained algorithms are needed to enable such automation. Machine Language and AI can enable these.
Where is RPA useful
The target space of RPA is where traditional (API-integration) automation is either expensive or not feasible. Even within this space, RPA is more suitable in the following scenarios:
- The user interfaces of all the target technologies should be stable or controlled
- The manual actions were all rule-based and being performed in specific sequence
- There is no intent to move or make changes to current processes in near future
Down-side of RPA
RPA has become a huge success because most core business software (ERP, CRM, ECM, BPM, etc.) have a huge number of gaps in how they support business processes. Due to these gaps, human workforce has been forced to find less efficient work-arounds like spreadsheets, macros, copy-paste data, etc. The work-arounds have a cascading effect as each new way of doing things will create more new ways and therefore more inefficiencies.
If RPA is used to automate all these inefficiencies, it definitely would save huge costs, but these costs should not have been there in the first place. Some RPA vendors recommend that their software be deployed after streamlining the process – but in reality, that would be possible only with significant investments and innovation in the core business software. This leads to a chicken and egg situation between RPA and innovation. RPA appears to be winning here for now (not sure whether it is the chicken or the egg though).
Future of Process Automation
Smart Digital Automation – the Immediate Future
The concept of Smart Digital Operations is already evolving with a host of collaborative technologies coming together to automate the enterprise. These are typically a combination of two or more of the following technologies: Optical Character Recognition (OCR), Business Process Management (BPM), Robotics Process Automation (RPA), Machine Learning (ML), Natural Language Processing (NLP), Analytics, Workforce Management, Capacity Planning, etc. Together they can not only work with all types of data, they also enable attended automation,
Radical Universal Automation – the Vision
Most medium and organizations have a set of common core enterprise technologies – typically the ERP, CRM, Payments, Content Management, Collaboration, etc. – that cover the end-to-end of 70% of their processes. But since these contain only 20% of the organisation’s structured data, it makes a great business case to connect all these core technologies through any of the Smart Digital Automation technologies. But that still adds to the cost every time there’s a change in the core technologies.
An ideal approach would beall technology vendors adopt Open Innovation. As organizations haven’t gained artificial intelligence yet, I hypothesise the next best scenario (still in idea stage – needs brainstorming).
A few groups of core technology vendors can partner with relevant Smart Digital Automation vendors to form a dynamic syndicate that innovate and provide ready-to-integrate connectors (a la Open Bank Project).These connectors will now internally be accessible to all the syndicate members to develop real automated solutions that will cater to 70% of their processes. Organizations will wholeheartedly embrace these automations as:
- They will be genuinely process-driven
- They will be agile and can be quickly implemented
- Impact on change to other technologies will also be known, as they will be part of this environment.
- Integrations will be API-based, which means any UI changes will not matter
- Over a period of time, it would be possible to bring a structure to all types of data because 100% of data will be revolving in this syndicate.
Most importantly, innovation becomes the focus – competition and work-around technology will take a back seat – to take humans and technology into the next level of co-existence.