Orica has positioned itself as an early adopter of generative AI within ServiceNow, with an aim to expand its deployment across IT operations (IT Ops) for greater efficiency. Starting with a focus on deflecting IT service desk inquiries to chatbots and self-service solutions, Orica is exploring the potential of generative AI across 35 use cases in IT Ops. Orica joined the early access program for Now Assist, a suite of generative AI tools designed to augment the ServiceNow platform, marking a significant step in its digital transformation journey.
The journey to this point took over two years. Orica’s ServiceNow instance, originally implemented a decade ago, had drifted away from its out-of-the-box functionality. To address this, Orica committed to reimplementing ServiceNow, undergoing a comprehensive 12-month re-platforming effort in collaboration with Thirdera, a ServiceNow partner owned by Cognizant. This “greenfield” reimplementation allowed Orica to introduce new capabilities and set the stage for a full-fledged AI adoption.
Since completing the reimplementation, Orica has launched a broad range of ServiceNow functionalities, from virtual agents to HR and legal lifecycle events, cloud discovery, event management, and multiple integrations. Bradley Hunt, Orica’s DevOps and regional apps manager, emphasized that these advancements have been foundational for the company’s AI journey, which began with early proof-of-concept (PoC) implementations.
The company’s first AI integrations used Now Assist to connect to an external large language model (LLM) via Azure OpenAI. These PoCs demonstrated the potential value of AI in IT service management (ITSM), validating Orica’s direction. When ServiceNow introduced its proprietary Now LLM, Orica pivoted to using this tool, taking advantage of ServiceNow’s accelerator program to swiftly deploy features like incident summarization, resolution note generation, and AI-driven search in just six weeks. Virtual agent capabilities followed two weeks later.
Orica’s initial results with these features were mixed. While summarization and resolution notes added convenience, they didn’t generate substantial cost savings. Surprisingly, summarizing live chats and AI search became user favorites. The AI search, for instance, simplifies inquiries such as password reset instructions, eliminating the need to navigate through multiple knowledge articles.
Having proven initial value, Orica engaged Thirdera to activate the remaining Now Assist skills, including an automatic knowledge generation skill, which turns incident data into knowledge articles. After just two months of using Now Assist in production, Hunt reported very positive outcomes. For example, the virtual agent’s deflection rate surged from 18% to 94%, with success rates doubling. These metrics indicate that users are increasingly relying on the virtual agent, and overall service desk workload has decreased.
Approximately 75% of these virtual agent interactions are initiated through Microsoft Teams, underscoring the integration’s role in Orica’s workflow. Though it’s too early to gauge the AI search’s deflection rates accurately, Now Assist has already contributed to a reduction in Orica’s mean time to resolution (MTTR), shaving off a full day across all incidents.
To monitor and refine its AI capabilities, Hunt’s team leverages a dedicated Now Assist dashboard to analyze usage by region, skill, and team over time. Orica plans to continue optimizing Now Assist, aiming for full functionality within the next six to twelve months. This timeline will also allow Orica to leverage Now Assist’s language support for its global operations, which span across 12 languages.
Looking ahead, Orica’s roadmap includes improving deflection rates from service desk agents to AI-driven solutions. Furthermore, Orica aims to extend Now Assist’s capabilities beyond ITSM into broader IT Ops functions. Potential applications include alert and event management, software asset management, and more comprehensive support across IT operations.