AWiB @ IBM: Putting AI in Practise – Streamling customer information management through chat bots and RPA

In Fall 2019, we were invited by AWiB corporate partner IBM to learn about addressing a business need by design and implementation of an AI application - in practise. Skills Academies focusing on “hard skills” or technology intensive topics have traditionally been popular among our network, and this event truly pushed participants outside of the comfort zone. Scroll down for a glimpse of the solution architecture of a simple yet efficient solution using a chatbot and Robotic Process Automation!
The evening was opened by IBM Finland CEO Mirva Antila. In Finland, IBM has 600 employees and a strong focus on diversity: for instance, the local Women Leadership Council is active in organizing events, mentoring and setting the employer brand into practice.
To put the AI case exercise into business context, IBM consultant Niklas Syyli introduced us to a recent customer case. Together with IBM, a Finnish bank had built an online chat service to match the time and place independent communication preferences of today’s customers. On average, millennials send an astounding 67 messages a day. The bank’s own customer service experts were involved in mapping the chat flows and training the AI to address a variety of situations. Overall, IBM is deploying bots for e.g. consultative sales and ecommerce, customer service, and internal information banks for corporations, to name a few use cases.

After the intro’s, IBM AI and automation technicians Sandra Calvo Martinez and Jukka Juselius helped us to get into action of building a simple AI solution by ourselves. Given the limited time, the exercise was more about configuring and connecting solution components rather than writing code from scratch. All in all it served as a great intro for business and technology backgrounded participants for components and work steps of forming an AI solution. 

Overview of the case exercise: building a simple AI solution consisting of customer action (1), chatbot (2), managed workflow by RPA bot (3), human verification (4) and automated workflow (5).

​Here is the case solution in a nutshell:

1. Customer has a discussion with a virtual agent (implemented with IBM Watson Assistant service) to inform about their need to change a contact address.
2. The virtual agent gathers the customer’s address information and their business ID and then starts a managed workflow (running on IBM Business Automation Workflow environment) to handle this new information.
3. First, the workflow triggers a RPA bot (implemented with IBM RPA) that opens public business ID register (YTJ) and makes a search using the business ID given by the customer. Bot then extracts the official address information for that specific business ID and sends the results back to the workflow.
4. When the bot has gathered the official address information from YTJ, workflow automatically moves forward to its next step. It brings up a task that a human needs to handle (verification of customer provided and YTJ information gathered by the bot). The handler can inspect if these match and can then decide to accept or decline the address information change for the customer.
5. The rest of the workflow is fully automated and executes based on the decision our handler made. If the change request was accepted, workflow would save new address information to our CRM system and notify the customer, but if not, just notify the customer of the situation.

 
In step 2, we got to test the intents and entities of the chatbot before connecting it to the managed workflow. To enable step 3, we got to teach the RPA bot the simple workflow of retrieving company information based on business ID. This part was especially instructive, as it demonstrated how manual workflows could be eliminated in practice. 

A glimpse under the hood for building a RPA application: using task recording and linking actions to HTML objects

 Many thanks for IBM’s organizing team for a challenging yet fun evening – this was Skills Academy at its best!

Movie tip: Hidden Figures (story about female NASA mathematicians who were the brains behind the launch into orbit)