Skip to Content

From digitization to algorithmisation: How chatbots combine RPA, AI, and ERP


This blog article is part of our RPA (Robotic Process Automation) blog series and focuses on the underlying opportunities of AI-enabled RPA in an ERP ecosystem. Throughout this article you will learn: how to overcome the boundaries and boost RPA capabilities, how chatbots can automate master data management, and how chatbots can be implemented in RPA and ERP systems.

Chatbots can simulate and automate human conversation through voice commands and text chats. They can be used through messaging applications or web pages and are naturally deployed in dialog systems for various application fields (e.g. customer service). AI enables smarter chatbots to understand unstructured human input by applying natural language processing (NLP).

How to further increase the potential of RPA and overcome limitations?

In the recent years, RPA has been one of the most impactful technologies in process automation across all types of organizations. Most of the RPA systems rely on structured data, static rules, and recurring events. RPA captures structured data based on the rules engine to perform the pre-defined workflows. However, the static setup does not allow processing of unstructured data. AI is the critical game changer that adds an intelligence layer on top of RPA systems so that the bots can handle unstructured data using their dynamic rule set. Once setup, RPA can manage exceptions, and the system can continuously improve itself with further training. AI can derive insights from unstructured data and deliver structured data to the existing RPA systems.

Algorithmisation is the process of gathering information from (unlabeled) data to use for Machine Learning and create new processes and data for additional processing. Chatbots that can be integrated into existing RPA & ERP ecosystems to capture structured data from human conversations and push into back-end systems.

How can Chatbots support Master Data Management?

Chatbots can further optimize the master data management process. Employees are often required to manually structure the unstructured customer inputs (e.g. email messages, phone transcripts) before the data can be processed by RPA technologies. A chatbot can automatically update the customer master data and support back-office employees to automatically enter the data by using automated communication channels, saving time for the customer and employee. This allow employees to focus on exception handling and value-add tasks vs. manual data entry.

Possible use cases range from updating names, postal or email addresses, phone numbers to more complex cases where more data needs to be analyzed by AI, like changing order details, contract data, or authentication methods. These use cases have several benefits in common:

  • Direct connection between different systems (with the same front-end)
  • Scalable AI and RPA capabilities (from simple to complex use cases)
  • Best of both: combining AI strength (unstructured data) and rule-based RPA
  • No human interaction required (reducing errors and time)

Using automation to update customer data from front-end to back-end.

How to setup a chatbot with RPA and ERP integration?

Over the course of eight weeks, we developed a tangible AI MVP (minimum viable product) for automatic master data updates, which used a chat to capture customer requests to update their mailing address. We utilized IBM Watson and its conversation-as-a-service for the NLP capabilities of the chatbot. It was entirely integrated into an existing UiPath and SAP system and also worked with other RPA and ERP software platforms.
Successful execution requires end-to-end expertise to understand the various tasks. Capgemini Group is delivering the Chatbot MVP with a team of experts across fields:

Consulting Services (Capgemini Invent):

  • Leading the overall program and providing expertise in RPA, ERP, and digital customer journeys
  • Use case creation, functional and business specifications / requirements, setup of UiPath and SAP workflow, and agile project management

Technology Services (Capgemini – IBM Watson Team):

  • The Capgemini APPS – IBM Watson Team is responsible for the development of the AI conversational service, core application, and the front-end development

Business Services (Capgemini):

  • Combine RPA and AI technologies and seamlessly integrate interfaces

Experience Design Services (Idean):

  • UI design and visualization of the chatbot characteristics

Additional support from other 3rd party software partners can also be considered based on the client’s specific requirements and needs.

Based on our experience, we suggest approximately four months to implement a chatbot into a live environment. The MVP was developed in two months, but integration to the live back-end system and training for the NLP capabilities required two additional months.

Starting with a hands-on, value-add use cases with feature-rich components should be considered for the use case pipeline. Some examples include: chatbots for customer complaint management (including predictive analytics), Machine Learning for customer credit assessment, or Deep Learning enabled image analysis and knowledge engineering for automatic handling of customer order entry.

What are the key considerations for adding AI to RPA processes?

  • Direct integration of chatbots with existing RPA & ERP systems provides both customer value and organizational benefits
  • AI is the new UI: Chatbot is the new user interface (UI) for customers to interact with companies in a faster and natural way
  • To maximize the value of smart automation, companies should utilize a holistic approach to process automation. Leveraging experts with the right skill sets and business, technology, data science backgrounds will support the success fo the program

Add an intelligence layer on your existing RPA and ERP systems to enable smart end-to-end automation and break through the boundaries of unstructured data. Start the journey from digitization to algorithmisation today!

Find out more about this topic:

Author detailsNA Contact details
Kai Broek
Wolfgang Enders