Inspired by the conversations we had with our customers recently we decided to look closely into the topic of chatbots and to showcase some use cases which might be of interest specifically for HR.
The idea of conversational platforms is obviously not a new one and a lot has been written on this topic. Progress in language models is creating new break-throughs (e.g. BERT or more recently OpenAI’s GTP-3) with the percentage of people being able to identify AI-generated news articles constantly decreasing in the recent years as quoted recently by the The Economist.
You will find several taxonomies of chatbots in the literature. Our claim is that even with scenarios which do not require language understanding you can reach interesting automation potentials, revamp your channel strategy within your business function and gain helpful insights from your organization in order to move up on the technology ladder.
Let’s look at our first example of a chatbot as a transactional self-service application. We created a sequential conversation for a well-know scenario of changing your address.
In the first step, using Microsoft Teams, the employee needs to login and identify themselves with the company's corporate identity provider.
Since SuccessFactors, being our backend in this scenario, defines several address types the employee is asked to select one they want to change.
Next, the employee is presented with a sequence of questions on the corresponding address elements, each step of the conversation providing the possibility to validate user’s input.
Finally, the employee is presented with the collected address information, the address is validated (point address, small corrections being conducted automatically) and if it exists it is stored in SuccessFactors.
Let’s highlight some features of this offering, though other technologies are obviously possible:
- We use Microsoft Teams and web as a channel
- We connect to SAP SuccessFactors as a backend to store employee data
- We integrate with Azure AD proving a user sign-in functionality
- We leverage Microsoft Bot Framework and
- Last but not least, we offer long-missed address validation functionality to improve the data quality in SuccessFactors
In the following parts of this blog series will focus on further scenarios, some of which require NLP and context information. In the mean time, reach out to us for a demo or if you have any questions on the topic of chatbots and conversational AI.