In the latest version of its Now platform, Xanadu, ServiceNow has taken the first step toward Agentic AI. In other words, AI agents that can perform automated actions based on behavior and actions of real employees. In doing so, ServiceNow is taking self-service to new heights.
We already wrote about the new version of the Now platform. Many versions of Now Assist have been added to the various ServiceNow applications. These can help the ServiceNow user summarize text, for example, from tickets and incident reports.
Agentic AI – AI agents
ServiceNow has introduced two AI agents, one for Customer Service Management and the other for IT Service Management. Organizations can also start building their own AI agents on the Now platform. In developing these agents, ServiceNow says the focus is human-centered. “It has to be a true collaboration between humans and AI,” ServiceNow said. That requires a bit more explanation. The idea is that an AI agent learns from a human what tasks and workflows it may perform, as well as how to perform them and when further human control is required.
Once the ServiceNow administrator is convinced that an AI agent can perform its tasks successfully and without error, human control can also be removed.
Developing skills and prompts
ServiceNow users can now develop skills and prompts and add them to an AI agent. This can be done with the Now Assist Skill Kit, which allows users to develop, test, and deploy the skills to specific AI agents. The skills can use generative AI and start workflows. For generative AI, the ServiceNow LLMs can be used, but models from third-party vendors or possibly even in-house-developed models can also be used.
ServiceNow users will develop a separate AI agent for all the tasks they want to automate. The idea is that the AI agent learns from human workers how to use these skills in practice. The different AI agents can be invoked through interaction via Now Assist.
In the future, ServiceNow will also add interaction via audio (your voice) and images to invoke an AI agent. This will allow you to have a conversation with an AI agent, for example.
Data is hugely important
Data is crucial for the success of this new strategy with AI agents. Earlier this year, we wrote about RaptorDB, ServiceNow’s new PostgreSQL-based database.
The RaptorDB edition presented earlier this year was more or less downgraded to the standard edition. There is now a new version, RaptorDB Pro, which is much faster than its predecessor. Most ServiceNow customers are still using MariaDB today but will move to RaptorDB in the next few years, which should give the ServiceNow environment more speed and better handling of AI. Customers who use AI heavily may be migrated faster.
ServiceNow continues to innovate with AI
ServiceNow is now taking a big step toward AI agents. In doing so, the company is innovating toward the next AI milestone. Over the past week, we have seen the entire industry move in that direction. The question, of course, is who can offer that the fastest in a good way. ServiceNow is one party to consider. AI agents will soon perform many self-service and support tasks until things get complex. Once you need level 3 support, you will still need a human.
The future will have to show how fast these AI agents can innovate to automate level 3 support as well. For basic tasks, we expect AI agents to dominate soon.
Also read: Salesforce raises the AI bar with Agentforce; are AI agents replacing employees?