Robots are here to stay, and are now firmly entrenched in organizations. However, once an organization has leveraged RPA bots to reduce the manual and repetitive workload of multiple functions and processes, limits usually appear when they try to deploy RPA more widely across the organization.
AI at the service of process automation
The first constraint of RPA concerns the data format for feeding a robot. Robots can’t directly handle heterogeneous input or unstructured formats (scanned documents, etc.). This usually leads to exclusion of certain processes from the automation field, and users must be tasked to manually reprocess the input data in order to feed the robot with structured data. This is not an ideal set up as it maintains low-value tasks for human employees.
The other common limit is the “cognitive” tasks that exist in many processes and which cannot be automated with RPA technology. These are tasks where the rules can’t be modeled and where experience of operational staff is required. Examples include interpretation of a request expressed in a mail, judgment on its priority, and decision-making on the continuation to give to a particular case.
RPA and AI – the possibilities of cognitive automation
To break the wall of complexity and automate cognitive tasks, it is necessary to mobilize AI technologies. Solutions available on the market provide ready-to-use AI services, such as optical character recognition (to transform a scanned printed document into a text document), transformation of voice in text (“speech to text”), and sentiment analysis, etc.
These solutions can be helpful for many processes, but are not always sufficient and sometimes need to be combined with more advanced solutions. For example, when the input documents of a process are of various natures and formats, it makes sense to apply machine learning solutions. Such solutions can automatically learn from examples given by users, drastically improving the time to market of automation compared to the classic development approach where all rules need to be coded.
AI hence offers new possibilities for automation by pushing the limits of RPA, delegating even more tasks to the machine and generating greater gains. The RPA solution landscape is evolving quickly either to better integrate with third-party AI solutions, or to directly embed these new capabilities. Organizations need to embrace this new opportunity in order to upscale their automation ambition and program by leveraging the potential of AI.
To learn more about how Capgemini’s RPA and AI solutions can deliver enhanced value for your organization, contact: firstname.lastname@example.org
Fabrice Perrier focuses on the impact of intelligent automation in banking and insurance. He supports clients in positioning and deploying such transformations, leveraging the potential of robotics and AI as well as more traditional levers, and ensures conditions for sustainable results by engaging business and IT in new and industrialized operating models.
Thank you to Alexis Jarroir, Artificial Intelligent Expert, Capgemini Invent, for his input to this article.