Case Study

A friendly AI-based chatbot for HR

Short, robotic answers and a lack of understanding contexts will soon be a thing of the past for the employees of a leading global technology company using the internal HR chatbot. A completely new technology base from Cluster Reply will ensure this in the future.

#Knowledge management
#Generative AI
#Chatbot 
#HR

Picture

The challenge 

Transforming an outdated chatbot into a state-of-the-art solution that utilizes the full potential of Generative AI.

The scenario

An update for the chatbot

Vacation, new company car, travel expense reimbursement: In 2017, a leading global technology company  introduced a chatbot to provide its employees with information on HR topics like these quickly and easily. A practical solution, but it is no longer state of the art. It often responds very briefly and robotically and is unable to relate several successive questions. A real conversation is therefore not possible with it. In addition, the chatbot requires a lot of manual maintenance when answers to certain topics need to be updated. To change this, the company relied on the expertise of the experts at Cluster Reply. 

Picture

The solution 

Contextualization, follow-up questions and a friendly tone 

Thanks to Generative AI, the chatbot now is  able to contextualize the employee's questions and thus conduct a real conversation. It is currently proving this as part of a minimum viable product project realized by Cluster Reply. The updated version of the chatbot suggests further questions if the question asked does not contain enough information to give a clear answer. If it is nevertheless unsure, it forwards the request to the HR department and thus avoids errors.

The chatbot also reveals whether it has used internal data to answer the question or whether the Internet was used as a source of information. This allows employees to classify the information and rely on it. The new concept is rounded off by a new friendly tone, which replaces the formerly factual and cool wording of the chatbot.  

How we did it

A new technical basis

To achieve all this, the experts at Cluster Reply gave the chatbot a completely new lease of life: 

Picture

Azure


The container apps, OpenAI Services, the Form Recognizer and the Search Services from Azure form the new technology stack. They enable a seamless experience and ensure a high standard of data security and processing efficiency.  

Picture

ChatGPT


Instead of a classic keyword search, the chatbot now uses the Large Language Model from Open AI. Thanks to this new technological basis, the chatbot can enable a conversational flow and recognize contexts. 

Picture

Advanced prompt engineering 


By developing and using customized commands that the chatbot submits to the language model in the background, the chatbot takes on a personality and responds in a friendly and professional tone.

Picture

Automated data pipeline


To reduce the manual maintenance effort, the experts at Cluster Reply have set up an automated data pipeline. This makes content management easier and supports different data formats. 

Picture

Teams


Until now, the chatbot could only be accessed via a web browser. Thanks to the new Teams integration, the employees of our customer will be able to ask their questions conveniently in the chat application that they already use on a daily basis. 

The result 

A success story will be expanded 

The new version of the chatbot is currently being tested as part of an MVP project, but our customer is working to finalize the deployment. After all, thanks to the chatbot's new technology base, employees benefit from easier and more convenient access to reliable information. They not only know the answer to their question, but also the source and can therefore rely on the information. And so the winning concept is likely to be rolled out in other areas in future.

Picture

Cluster Reply is the Reply Group company specializing in consulting and system integration of Microsoft technologies. As a Microsoft partner, Cluster Reply is active in Germany, Austria and Switzerland and works within the Reply network with sister companies in Brazil, the UK, Italy and the USA. The company focuses on innovation and supports customers in their digital transformation. The solutions range from on-premises to cloud applications in the areas of modern workplace and security, business applications, applications and infrastructure as well as data and artificial intelligence