Robotic process automation (RPA) software is the fastest-growing segment of the global enterprise software market. It’s easy to see why. Intelligent Automation (IA) and RPA tools automate repetitive and mundane tasks, freeing up employees to do more high value work. In turn, this helps organisations deliver better customer experiences, increase business agility and improve productivity. Yet, many companies are struggling to realise these benefits, or are not exploring them at all, due to the real and perceived challenges associated with the technology.
Although IA and RPA have applications in almost every industry, many of the logistical and technological challenges that businesses face, from making the business case to automate, to implementing and embedding the technology, are the same.
Achieve scale in IA
IA has many individual sensory capabilities; from using image recognition to scan photographs, to converting the spoken word to text, to predicting the future based on past actions. However, scaling this technology is essential to achieve true business transformation. Getting it right means marrying vision and strategy. This follows three stages:
- Establishing capability: In this phase, big picture thinking is needed to define and communicate a shared vision and create structured organisational roles – as well as tending to the details of governance boards; agreeing on a standardised approach and communicating outcomes against KPIs. Personnel should also be a key consideration: with questions around training core teams and establishing delivery methodology key to this part of the process.
- Replicate & ramp up: Once this vision is established it’s about replicating and ramping up the processes underpinning it. RPA needs to be aligned with future organisational design decisions, with examples of success used to incentivise staff to buy-in and encourage seeking out new automation opportunities. Core teams can be called upon to train and mentor new team members and establish codes of best practice.
- Deliver differentiated performance: The final and ongoing stage in the process is delivering differentiated performance. A virtual workforce should be embedded at the heart of the organisation with a seamless handover of work between humans and robots while minimising disruption.
Security by design in IA
With any technology, security absolutely needs to be a main priority. For IA the biggest security issue often arises at the point where human and machine interact. For example, human error during an automated financial reporting process can result in losing-man weeks – not days – and a delay on the reporting of the group’s finances.
Other security issues that need to be considered include rogue access; data loss; hacking; privilege abuse; vulnerabilities and malware, which all show the centrality of security to IA implementations.
But like well backed up data – all is not lost! Those looking to implement IA should heed security protocols like encrypting data and multiple layers of authentication, along with reducing access rights and requiring human validation on certain processes.
The unforeseen benefits of IA
No discussion of IA would be complete without an examination of the organisational changes that are taking place. New technologies will create a number of different roles in the future.
When we think of the benefits of automation, we typically consider things such as time saving, headcount reduction and reducing processing times. But there are other significant benefits which don’t typically appear in business cases:
- An empowered workforce: removing menial tasks from your employees would pave the way for switching employees to more value-adding work. This consequently builds capacity and drives larger financial returns over the long term.
- Standardised, yet agile: automation creates a double win by not only standardising process execution but also allowing for rapid changes to the standardised processes to cater to business requirements and end-user demands.
- The process improvement opportunity: introducing automation within an organisation is an opportunity to understand the business need and potential added value processes can deliver. This is crucial as they are seldom designed, documented or revisited.
Overcoming roadblocks to successful AI implementations
IA impacts also AI. Faulty data or human error could affect AI and its smooth running. IA can overcome the roadblocks in AI implementation.”.
AI “eats data”. It needs a lot of correct data to be able to function. There are practical considerations – things like enforcing two factor authentication – but human considerations as well. Indeed, successfully implementing AI also meant incorporating fairness; reliability & safety; privacy and security; transparency and accountability.
What AI is not
It’s also important to remember what AI is not. AI, contrary to popular belief, does not see output simply improve over time. Indeed, in the beginning, AI functionality is based on what it has been taught by humans. As soon as it is in use, it collects more data and thus becomes more and more precise.
But before AI is used, it is trained with pre-selected data. As soon as it stops being fed with training data, AI begins to classify new data in the same way it has been trained before. But AI does not learn anything new. If AI cannot categorise some of this new data, the accuracy of the output worsens (called “data drift”). In which case, AI has to be retrained.
This goes hand in hand with one of the biggest misconceptions of AI projects. The work does not stop after the initial roll out. AI models have to be continuously verified while they are in use.
And, as always, we must think about money too. While some form of AI applications do require the work of expensive data scientists and/or computational linguistics, an increasing number of AI software tools are becoming easily accessible to businesses and do not require big investments. While advanced AI technology does require deep understanding in programming languages, most enterprises will opt to leverage business applications developed on top of tools that were built by companies such as Google, Amazon or expert startups. Examples are Amazon’s Alexa or Google Home which provide multi-lingual voice recognition. By using these existing tools, the business value lies in configuring customer centric components tailored for specific needs and less in the application of data science.
IA and AI has already been shown to have profound and far-reaching benefits and, while it is difficult if not impossible to predict with complete certainty where the technology will go next, what is more certain is that its use will only proliferate.