Seidor
automatización

22 October 2021

Chatbot, RPA and artificial intelligence: the triple threat for true process automation

To achieve success in process automation, it is essential to cover the entire end-to-end cycle. From people (customers, suppliers, workers, etc.) to every single procedure included in a specific workflow. In this article we go over how we can guarantee a quality process and provide users with an optimal experience thanks to the combination of three different yet complementary technologies.

When we depict the different layers involved in a process, from interaction to optimisation in SAP, we see three complementary technologies:

  • Conversational Artificial Intelligence (CAI): This is the interaction tool, or the interface through which people communicate with what is known as a chatbot: software that manages the conversation in order to obtain the information necessary to successfully execute the process.
  • Robotic Process Automation (RPA): This is the execution tool that performs the various tasks automatically, either through APIs, services or screen simulation (SAP, other programs or websites), by filling in fields and clicking buttons.
  • Machine Learning (ML): This is based on a set of algorithms that provide the necessary intelligence to optimise data and extract conclusions as similar as possible to what a person dedicated to the process would produce.

These three components maintain continuous interaction and a constant exchange of information in the same way that a person involved in the process would do. For example, to provide adequate responses during the conversation, it is usually necessary to harness information from different programs (RPA) or even activate different machine learning algorithms to better understand what the user is saying and to offer accurate and quality responses.

Similarly, during execution with RPA, we may need more information from the user, which makes it necessary to re-launch the conversation (CAI) or activate ML algorithms to infer information based on how the process has been previously solved or through a probability calculation.

In short, this is a process of continuous interaction, execution and optimisation until it is completed without errors and to the user's full satisfaction. With all this in mind, the maturity of the market becomes a compelling reason to adopt this suite of technologies.

Future prospects: Conversational intelligence as a key element for companies

At SEIDOR, we have found that developing individual projects is no longer an option. Success depends on the adoption of a measured and controlled strategic approach that allows for scalability across languages, channels and the company itself. For this, the power and versatility offered by the latest technologies are essential.

Looking towards the future, we expected that a customer will be able to handle most of their relationship with a company without human interaction. Against this backdrop, conversational intelligence, even voice activated (integrating devices like Alexa at company level), will play a key role. However, there are still many challenges to be resolved, and we have been working on them for some time at Seidor.

"It is essential to apply a global perspective with a focus on digital transformation"

In parallel, integration via an intranet of conversational AI applications is another crucial component. This smart routing will allow the cross-application handover process to be carried out in several ways. One of which includes using a master application or superbot capable of guiding and delivering various processes in a fully integrated manner. To do this, it is essential to apply a global perspective with a focus on digital transformation, and, at Seidor, we deploy the appropriate strategies for correct implementation.

Technological challenges to move towards maximum integration

All this analysis has enabled us to identify the main challenges facing each of the technologies:

  • Improving the customer's chatbot experience, as it should not only answer their questions, but also speak, think and, above all, develop emotional relationships with users by performing an appropriate analysis of feelings.
  • Applying the latest developments in natural language processing, as well as unsupervised learning and reinforced learning algorithms, to provide chatbots with sophisticated algorithms that allow them to bring unique and customised experiences and generate more authentic relationships with a specific target audience.
  • Shortening the time required to implement the bot with sufficient quality and without the need for long simulation periods. This involves using previously trained bots (BERT, ULMFit, etc.).
  • Accelerating processes that are most likely to be automated in the SAP product ecosystem, both from the point of view of automation in RPA and the intelligence applied for its optimisation.
  • Integrating other technologies that can enrich automated processes, such as blockchain for processes involving customers, consumers or suppliers.

Finally, it should be noted that, at specific moments during an interaction, there is no substitute for the empathy that human agents can offer or the creative intelligence needed to solve a case. In complex or unusual circumstances, it is often the human capacity to establish parallels with similar experiences that solves a problem.

In short, chatbots must be able to interact seamlessly with a human when necessary while ensuring that all the information gathered is transferred so that the customer does not have to start again, as this can cause frustration.