Twice a year, a new version of the ServiceNow Platform is released. Now, it’s the turn of Xanadu, ServiceNow’s major AI release. The update adds hundreds of AI features. Only a tiny part of all the new AI features were shown during a presentation, but ServiceNow also came up with its vision for the future: the company wants to bet on Agentic AI.
At ServiceNow, they see AI as the tool to take the Now platform to the next level. That’s certainly true, the Now platform benefits tremendously from AI. Users of the Now platform perform an enormous number of actions and workflows on the platform. Many of those actions can be faster and more efficient if you add AI to them. You can also automate a lot more
ServiceNow makes it too complicated
However, the trick is keeping AI understandable for all ServiceNow users. Given the hundreds of new AI features that ServiceNow presents today, you quickly get lost in all the different AI features. In addition, American software companies tend to make it all a bit too complicated. If you zoom out to what ServiceNow is adding to its platform today, those hundreds of AI features are a lot of the same thing.
Most features often boil down to just two use cases.
Summarize and generate text
With many new AI features in the Now platform, it comes down to text summarization and generation. For example, the platform can grab all the interactions within a support ticket and generate a summary. You can also replace a support ticket here with a change request, incident, etc. In addition, it can generate responses to tickets, chats, and emails using a short command. For example, “tell the customer that this is not possible.” An elegant email with a friendly text will be generated to inform the customer that the request cannot be fulfilled.
Now Assist, the intelligent chatbot
ServiceNow has developed an intelligent chatbot with Now Assist that was initially available in a few solutions. Now the chatbot is coming to many more ServiceNow applications. Some examples; Now Assist for Security Operations, Now Assist for Sourcing and Procurement Operations, Now Assist for IT Service Management, Now Assist for HR Service Delivery, Now Assist for ITOM and Now Assist for CSM.
What Now Assist offers is an intelligent chatbot within all of these applications where the user can ask questions using NLP (natural language) and get help to perform tasks. These can be simple questions. For example, what is the policy for requesting a new laptop? Now Assist can answer based on existing knowledge base articles. It can also be a question followed by an action. For example, in an HR application; how do I request personal time off? The Now Assist will understand and ask the user when it would like to have PTO and put the request directly in the HR system.
However, it can also be much more complex. Take, for example, Now Assist for Security Operations. It supports summarizing incidents and helps prioritize incidents through a set of questions.
Where other platforms choose to work with a single CoPilot, ServiceNow opts for ‘Now Assist for ….’. On the dot in this case comes the concept for which the chatbot is intended, for example ITSM or CSM. By doing so, it wants to emphasize that each Now Assist is also trained for a specific application. For example, Now Assist is often trained with knowledge base articles and past questions, cases, and tickets and how they were resolved. It varies by ServiceNow applications, how intelligent Now Assist is, and how much help it can provide. For example, for a ticket, Now Assist for ITSM can also show similar requests and how they were resolved. This also applies to customer service.
Build your own workflows to improve or expand Now Assist
A ServiceNow administrator can also build workflows and link them to common queries and actions. To cite the earlier example of the laptop policy, suppose the policy is that an employee may order a new laptop every three years. Now Assist could also have started a workflow that checks in the system how old the employee’s laptop is. If it is older than three years, it could provide the option to go through a process within Now Assist to order a new one.
Now Assist can ask what kind of model the employee wants based on the company policy, whether the address information in the HR database is still current and then automate the order and delivery.
Now Assist Skill Kit
Summarizing text or tickets within a ServiceNow application is a skill. A skill developed by ServiceNow that defines what the platform and an LLM should do. It may involve grabbing all interactions of a ticket and creating a short summary. With the Now Assist Skill Kit, ServiceNow users can now develop their own skills. Users can then use all the data and knowledge within the ServiceNow platform, as well as define the context and what kind of response they require from an LLM. This can be done based on the standard LLMs offered by ServiceNow, but also by linking their own LLMs.
It can generate fully customized responses that allude to specific situations. It can for example output data in a specific format.
Workflow Studio
The previous ServiceNow Washington release included the new Workflow Studio application, which allows users to build their own workflows using AI. When you combine that with Now Assist and the Skill Kit, very sophisticated tasks can be automated. If you embed an additional confirmation query into a workflow, you can automate many things. ServiceNow wants to facilitate this in an even more advanced way in the future.
ServiceNow wants to integrate Agentic AI into the Now platform
Agentic AI is a term that has come up more often lately. It is the next step in artificial intelligence and ServiceNow can benefit greatly if they can roll it out successfully. Whereas generative AI generates context like text and images, Agentic AI focuses on making decisions and executing actions.
Agentic AI can make and execute decisions without human intervention. Now Assist can currently help users submit tickets or, on the contrary, help process tickets. Also, in some cases, it is already possible to show a human agent what is likely to be the solution for a particular problem. This is done based on similar past cases. With Agentic AI, the latter human action could also be automated. However, the AI must be sure that the chosen solution is the right one.
That will be ServiceNow’s biggest challenge. When can you deploy Agentic AI, and how do you determine if the chosen action is also the right one? ServiceNow has not yet mentioned a timeline for this, but it is certainly interesting. It also opens the door for new discussions because the AI will make decisions that might get disputed. For example, will organisations do fraud detection with Agentic AI? That means an AI can tag someone as a fraudster. It’s too early for that discussion since Agentic AI exists only in theory for now, but it’s good to think about it.
ServiceNow Xanadu is now available
The latest version of the Now platform is now available. In ServiceNow Xanadu, customers have many new AI features to get started with. It is now up to organizations to identify the AI features that can make a difference for their workflows and business processes. We are especially curious to see how quickly ServiceNow customers will embrace and use these AI features. Something ServiceNow will no doubt tell us next year.