Generative AI is potentially the most transformative new technology since the introduction of the public internet, and it already has many exciting applications within enterprise service management (ESM). AI is known for enabling intelligent chatbots, predictive capabilities in ticket management, and the ability to identify emerging service issues long before they become problems. Generative AI promises an entirely new level of innovation.
With generative AI, ESM platforms could not only understand the context of a user request in natural language but also comb through multiple knowledge bases to provide much more than links to information. Generative AI could deliver a clear solution that summarizes the most relevant and current information across multiple languages.
In fact, it could go well beyond simply providing information — it could automate the solution. If numerous employees are logging tickets about problems logging into the VPN, HelixGPT can analyze that data, determine a restart of the VPN is required, and kick that process off automatically. And regarding those knowledge bases, generative AI could automate much of their maintenance, eliminating outdated content and incorporating new information.
Despite the intense hype around ChatGPT, that instance of large language model (LLM) technology would not be appropriate for enterprise use. One of the biggest challenges with ChatGPT is controlling its output. Organizations want to make sure that the generative AI properly understands the tasks at hand to prevent incorrect responses and limit “hallucinations” or false information.
BMC Software addresses these limitations with HelixGPT, which enables the world described above through AI Service Management. LLMs like ChatGPT are built from publicly available knowledge, but enterprises operate within a specific domain and need generative AI to prioritize enterprise knowledge over world knowledge. Each company, for example, has its own policy for employee password renewal, making it difficult to nail down a universal standard.
Additionally, BMC enables HelixGPT to tune itself to enterprise-specific information with additional layers that prevent hallucinations. Of course, HelixGPT will not be limited to enterprise-specific information — if the query requires outside, public information, it will be able to draw on it.
“When it comes to generative AI and ESM, we see the potential for innovations everywhere,” said Ajoy Kumar, Chief AI Scientist, Service Management at BMC Software. “Users will see better search, better answers, and a more human-like virtual agent. And they won’t just get pointers to solutions. HelixGPT will provide action.”
