ChatGPT is the online technology that has seen the quickest adoption, as measured by the time it takes for new technology to reach 1 million users, according to Statista. It is therefore no wonder that many vendors are looking at the benefits of Large Language Models (LLM) such as OpenAI GPT-4 to the ITSM domain.

OpenText has been a pioneer in the field of applying AI/ML technologies to ITSM with SMAX (actually, its predecessor, HP ServiceAnywhere) being the first AI/ML-powered service desk in the industry, and has since then expanded the use of AI/ML to many ITSM areas, the last one being for skills-based assignment of incidents. It has therefore been a natural expansion to augment SMAX with LLM.

ChatGPT integration

ChatGPT is great at bringing generative conversational capabilities, which can be used by service desk agents to search for publicly available knowledge and generate a response with good conversational quality.

The good news is that SMAX is ready to integrate ChatGPT: With the generic REST connector of SMAX Integration Engine, it is easy to develop a dedicated ChatGPT connector: ChatGPT can be added as a specific usage pattern in agent-to-agent chats, thereby turning ChatGPT into a virtual SMAX service desk agent providing well-crafted conversational answers based on publicly available information.

ChatGPT connector built on SMAX Integration Engine 

ChatGPT and Enterprise data privacy concerns

The flip side of the coin of ChatGPT's success is the concern about the privacy and security of data fed to a public LLM, regardless of the quality of its outcome. There have already been bans on ChatGPT in many large enterprises and most public organizations are forbidden to use it, sometimes as extreme as a country-level ban, such as in the case of Italy.

Even if ChatGPT improves its handling of data with respect to privacy, such reactions reflect a widespread concern about the use of LLM technologies for non-publicly available data.

OpenText plans on LLM and GPT

This is why we are planning to release later this year an implementation of our SMAX Virtual Agent chatbot with LLM-based capabilities but based on a private LLM, which will then combine the conversational power of Large Language Models with the privacy imperatives of enterprises and government organizations.

We will soon be sharing our detailed plans about that, so stay tuned…

Jacques Conand
Sr Product Management Director, Enterprise Service Management
OpenText