Generative AI comes with a set of challenges for the workplace—privacy, security, data integrity, and integration with existing processes. That’s why we’re building a private large language model (LLM) for enterprise use.
In my previous blog, What It Takes to Build a Game-Changing, Enterprise-Ready, Generative AI Virtual Agent, I shared how we’re building our private LLM. In a nutshell, we have four design pillars. First, hosted and managed by OpenText, our LLM is private and secure. Second, it uses enterprise data to provide relevant, near real-time answers. Third, it comes with access control so that employees engaging with the generative AI virtual agent may get different answers depending on their role, location, or department. Fourth, it will be available to assist employees whenever and wherever they need it. We’re exploring various ways of making the generative AI service ubiquitous—for example, by way of the ITSM self-service portal, embedded widgets in applications, public APIs, and integration with communication tools such as Microsoft Teams.
Without compromising enterprise data privacy, the OpenText generative AI virtual agent can deliver human-like, contextually relevant responses. Let’s explore the broad range of generative AI use cases for ITSM.
Assisting support agents
In this example, a support agent asks the OpenText generative AI virtual agent for advice about working with a difficult user. The virtual agent offers general tips that include listening carefully to the user’s concerns, offering tailored solutions to the user’s needs, and taking responsibility for any mistakes that may have occurred. The virtual agent also shares articles that the support agent might find useful. Article topics—such as Handling Difficult Calls, Let Me Talk to Your Supervisor, and Words to Use and Avoid—are semantically related (not keyword matched) to the support agent’s question.
Summarizing enterprise knowledge
The generative AI virtual agent can deal with complex topics that involve large enterprise documents. In this example, a procurement specialist wants a quick summary of a customer contract. After receiving the summary, the specialist asks some follow-up questions: What is the value of the contract? When will it expire? Are there any risks involved?
For that last question, the virtual agent analyzes and interprets the contractual risks with all the associated nuances and subtleties.
Answering frequently asked support questions
The generative AI virtual agent excels at answering typical questions related to IT (e.g., PC, email, connectivity) and HR (e.g., standards of business conduct, local company holidays), which lessens the burden on support agents. Answers to these questions are often drawn from real-time access to the latest enterprise news and updates. In the example below, the virtual agent fields a user question about a web mail outage and its expected duration. The virtual agent also offers options for accessing email during the outage.
The generative AI virtual agent can also answer how-to questions. In the example below, an employee wants to learn more about OpenText Core Content, a cloud-based content management solution. The virtual agent provides contextually relevant answers, drawing on product documentation as the knowledge base.
Automating service requests
Now let's look at a final example where the generative AI virtual agent automates a service request. An engineer in the DevOps team needs a new VM but isn’t familiar with the process of requesting one. The virtual agent readily submits a SMAX ticket for provisioning a new VM. The engineer can track the status of the ticket and will receive an email notification when the request is fulfilled.
The generative AI virtual agent capability will be available as a tech preview soon. Stay tuned for more updates!