The challenges around implementing intelligent automation are more often related to skills, leadership and data availability rather than tools and technologies.
Intelligent automation is one of the most important tools to increase company efficiency and resiliency. Its significance will grow massively in the coming years, surpassing traditional rule-based RPA. The relevance of leaders understanding opportunities around the marriage of machine learning and RPA is clear to see, as Gartner lists it as its #1 strategic tech trend for 2020.
But the transition will not come without challenges. When you work in technology, you get used to hearing familiar things. There’s always talk about robots replacing workers, automation going wrong, what happens in a power cut, and so on. The truth is, technology scares a lot of people. Yet, of the many challenges technology such as intelligent automation presents, the least mentioned are those relating to people. You only have to look at how much money is wasted on trying to transform a business using technology and you can see the scale of the challenge businesses face.
The challenges around implementing intelligent automation are more often related to skills, leadership and data availability rather than tools and technologies. This is something touched on by Chris Skinner in the article linked above, where “digital transformation is more about culture and organisational change than about technology”. So, let’s look at what I think are the three biggest challenges around implementing intelligent automation at the company level and dispel a few misconceptions.
Identify opportunities fast
Going beyond traditional process automation enables companies to automate far more complex processes and automate larger parts of those processes. Implementing AI within automation allows workflows to cover probabilistic decisions that traditionally have been reserved for knowledge workers only. This presents so many opportunities to take intelligent automation further so that it directly influences the day-to-day running of a business.
However, you have to recognise an opportunity before you can take it. Organisations often work in silos and don’t communicate well with one another. People who own a process ripe for automation sometimes expect too much from AI or they just can’t figure out what’s possible. Both can be attributed to simply not knowing enough about machine learning capabilities. And those who oversee the technical side of implementing automation processes sometimes don’t have enough data to work with, nor the right questions to ask. This leads to a disconnect between people that more often leads to scope creep or project abandonment.
Having a nimble tech stack that enables quick experimentations helps to identify the right opportunities from the sea of opportunities available. This is instead of starting a development project and, after months of work, figuring out if things are going to work at all.
Find the easy tools
Way too often, machine learning is an expensive add-on that comes with platforms like UiPath, Blue Prism and ServiceNow. And even with the bucks invested in the software licenses, it is hard to get the bang easily. Users still need to understand the ML models in order to work through the many steps required to get the automation into production. Yet, the platforms lack flexibility and customisability for intelligent automation.
Combine current platform capabilities with scope creep and you may end up abandoning projects, even when the actual project hasn’t been started. It can also lead to project completion, but at the cost of having implemented technology that’s old before you’ve even launched it.
Machine learning is changing all the time. Intelligent automation technology is moving fast and new developments are seen frequently. If a business finds itself down an expensive path of digital transformation using the wrong tools for the job, it’s a long way back. Not only can a poor decision over what tools to use be disheartening and expensive, but it can also lead to the questioning of intelligent automation as a useful option. Read more about the return on investment for Machine Learning.
Tools that are easier to use because they are built for normal people, as well as IT departments, are much better. We talked about how low-code apps and RPA will transform your business in a recent blog post, where the spotlight is very much on simplifying user interfaces so that automation can be enjoyed by virtually anybody. Imagine having lots of automated processes running across various business areas that are simple to update, simple to adjust and simple to use. It’s better than relying on cumbersome tech implementations that weren’t set up properly in the first place, and are already outdated.
Take care of the data
Like a pond full of expensive Koi Carp, machine learning models need to be fed correctly. The fish need perfect water, high-quality food and the right care. Intelligent automation needs good data. The problem lies in businesses having disconnected data spread across discrete platforms that are only available to specific groups of people, with limited access. This uncatalogued and unstructured data is incredibly difficult to organise and represents one of the biggest hurdles in trying to get an intelligent automation project off the ground. After all, this is the fuel for training the machine learning models.
As with our second example (above), lightweight low-code tools offer a solution to mastering unstructured data. By iterating scenarios using small steps (and small tools), a business is better able to evaluate the potential impact of the automation opportunity quickly. This helps to avoid common challenges while connecting those people most relevant to the process (and the data) with the use case early on. By adopting a centre-of-excellence-style dialogue between different teams across an organisation, the efficacy of automation can take hold. The whole workforce will become familiar with the benefits of intelligent automation and how it can help the business, promoting a culture of knowledge sharing around what IA can achieve.
Imagine the outcome of this sort of knowledge sharing. It could lead to more projects being delivered with less money being spent, and a workforce that develops and maintains automated solutions quickly and easily using low-code tools. Before you know it, the biggest human challenges around automation could turn into major cultural and technological advantages. The goal should be to find data-focused ways of recognising those automations where intelligence is necessary.
Effective management of the entire automation project helps to avoid such challenges.
Indeed! The right kind of process management, data management and tech/tools management, with a twist of the right type of leadership for your team, should do (most of) the trick.
Do you think effective management would also help determine fast what tasks/workflows/processes to automate, i.e. where ML+RPA is usable and where not? A challenge I face every day when talking with customers and industry experts.