Digitalisation is becoming more and more prevalent in the everyday operations of companies. Numerous administrative and business processes can now be digitalised and automated. Innovative modern technologies enable new fields of work and digital business models. To implement these digital business models, operational activities must run automatically and efficiently. Manual tasks must always be completed quickly and without errors so that employees have more time for strategic and value-adding tasks. This also applies to working on the screen: Less effort, shorter processing times, process optimisation, higher quality and lower costs can be achieved with Robotic Process Automation (RPA) and the use of Artificial Intelligence (AI), for example. Intelligent automation tools using RPA or AI are often used for the same goals, but their fields of application, strengths and weaknesses differ significantly. What are the reasons for procurement to choose AI or RPA?
What is RPA?
Robotic Process Automation (RPA) refers to the automated handling of business processes by software robots. These behave like humans: they perform rule-based work, start applications, select tasks with a mouse click and fill in data fields. In doing so, software robots take over especially data-intensive and repetitive tasks. After the software has been installed, other programmes such as email programmes, Excel or SAP are operated as if by magic. RPA is best used for activities that meet certain criteria: They should be PC-based activities that are data-intensive, repetitive, simple, non-value-adding, but error-prone. Furthermore, they should involve a high number of operations that are relevant 24/7. RPA works best with standardised processes that are started electronically and when the process flow changes frequently between IT systems. RPA is used in many business areas, such as: purchasing, logistics, customer service, warehousing, human resources, financial accounting, IT services, sales as well as production planning. Major industries for RPA implementation include banking, utilities, healthcare, insurance and telecommunications.
What is AI?
Artificial intelligence (AI) is a branch of computer science that aims to create intelligent machines by simulating and imitating human decision-making behaviour through various algorithms. Generally, AI is spoken of when a computer solves sophisticated problems in a single way that actually require the intelligence of a human being. Just like RPA, AI is used to automate and digitise manual, repetitive, simple but error-prone and time-consuming tasks. Wherever operational activities predominate and take time away from value-adding, strategic and creative work, AI technology can be used profitably. Operational procurement, for example, is predestined for this. The reason: here, a predominantly manual process chain with many error-prone activities still prevails. These activities are often still carried out “by hand”, such as receiving and checking supplier self-disclosures, sending enquiries to suppliers, obtaining quotations, sending orders, manually entering incoming customer orders etc. Similar to RPA, AI-based software is also used in numerous business areas and industries. Especially high-tech application fields of the future work with AI: search engines, machine translation, handwriting, speech, image and face recognition, self-driving cars as well as humanoid robots.
The areas of application as well as the advantages of RPA, such as relieving employees of boring and time-consuming manual tasks, time and cost savings as well as higher quality, are therefore obvious. But what about disadvantages or weaknesses?
Disadvantages of RPA
Complex configuration of RPA
While the configuration of RPA can be done by employees without deep programming knowledge within the relevant business processes, good business process knowledge is required. The software robot must be trained by the service provider just like humans. RPA can only be used where processes are already digitised. Thus, a process consultant is useful to evaluate which processes or components can be automated at all.
Lack of flexibility
Software robots are extremely inflexible. The platforms on which RPA interacts change frequently and the required flexibility often cannot be configured. Software robots are not able to react flexibly to deviations or variants, for example when process input data is not clearly structured or decisions within the process are complex and non-linear. Changing regulations that require minimal changes to an input screen can set back an RPA project by months. In fact, the technology may not work or work optimally when changes are made to interfaces, data or other inputs.
The more applications a software robot serves, the more challenging the scalability issue becomes. Of course, application systems are also regularly maintained, which can mean adapting the software robot each time, because software updates can change interfaces or field names. This limits the application breadth of RPA and also leads to a lot of additional work.
Monitoring for possible errors is necessary
Robots are not perfect either: employees have to watch out from time to time that the software robot does not make any mistakes. For example, if OCR (Optical Character Recognition) is used to read data from a PDF file, the reading process can be faulty. Since RPA technologies are not able to recognise errors or check data readouts for plausibility, it can happen that these errors are then transferred unnoticed to a database.
RPA is not worthwhile for every company
RPA is generally cheap to implement and operate. In some cases, however, it is more expensive and more time-consuming for a company than having the processes carried out by a human worker. This is the case if there are only a few standard processes to be processed in the company, for example because the company is very small.
RPA requires clean data
RPA requires a certain data quality. The effort required for its preparation and cleansing is often underestimated. Robots cannot do anything with ambiguous documents either. Moreover, RPA only works with input that can be digitised. Consequently, software robots can currently only be used for structured processes and routine activities. For semi-structured and unstructured processes, RPA offers little to no added value.
The solution for purchasing: AI-based software
If we look at the disadvantages and limitations of RPA, it becomes apparent that the benefits are rather limited, especially for purchasing. This is because companies do not always have a high (master) data quality from the very beginning. However, good data quality is a mandatory prerequisite for the successful digitalisation of business processes and models and the automation of workflows. Furthermore, for the use of RPA, processes in the company must already run completely digitally. However, this is seldom the case, even in corporate groups, and certainly not in small and medium-sized enterprises (SME).
But there is good news: there are already market-ready software solutions for procurement that offer real added value thanks to innovative AI technology. By automating operational processes as well as data and document exchange, business processes are sustainably standardised and optimised while master data is cleansed. In addition, some AI-based software solutions digitise all processes and eliminate media discontinuities – here, the optimisation process in procurement and the entire company starts much earlier and more comprehensively.
Another very important advantage of AI for purchasing is that it can support the exchange of information and at the same time relate content. During processing, it imitates the human behaviour of those involved in the process. In operational procurement, this applies to the automated extraction of data, the validation of information, the comparison of data and information, and the further processing or transfer of the correct information to the downstream company systems.
If AI is combined with cloud solutions, for example, unstructured data in free texts or documents such as order confirmations and customer orders can also be processed – unlike with RPA. Deviating terms in supplier documents, ambiguous units of measurement or rare languages are no longer a problem, because – another difference to RPA – AI has the ability to learn continuously. It becomes more and more precise over time, because the learning effect increases proportionally with the volume of data fed into it.
Furthermore, AI scores with high flexibility. AI has the ability to automate processes without clear rules and of high complexity. It can be used in many ways, reacts flexibly to variants and deviations; updates and adjustments to the application systems are no problem. This basic flexibility also goes hand in hand with easy scalability. This is of crucial importance, as procurement 4.0 in particular needs to be flexible, agile and efficient.
Summary: This is why procurement should rely on AI
- AI can also work with and structure messy data
- AI can solve challenging problems, understand complex processes and make decisions
- Very high data accuracy, hardly any errors
- AI supports and enables the digitisation of business processes
- AI is capable of learning and is becoming more and more precise
- AI is flexible
- AI-based applications are easily scalable
- High ease of use
Conclusion and recommendation
If you want rule-based, simple and repetitive tasks to be performed automatically, you can easily turn to RPA. However, in direct comparison with AI, RPA falls short of what is now possible. If you want a complete and future-oriented automation solution for procurement, you would have to combine AI and RPA. However, this is not necessary, as many AI-based intelligent automation solutions, such as the Netfira solution, offer a complete package and flexibly and reliably automate both simple and more demanding tasks and processes in procurement. Therefore, when choosing your software solution, look for flexible modules, high scalability, fast and high ROI, uncomplicated configuration and ease of use for the users. Furthermore, it is advantageous if the smart automation solution simultaneously improves the master data quality and digitised processes are not a prerequisite for the success of the tool, but a result of its use. Only in this way can automation solutions lead to the hoped-for process optimisations and increases in efficiency and pave the way for the procurement of the future.