Ever since the release of ChatGPT in 2022, the general use of AI technology has been a hot topic. The influence of AI has been present in our everyday lives in a subtle way, but now it is becoming a more obvious part of both our personal and professional lives. From human-like text chats to AI-generated images, everybody can now take advantage of AI on their computers, or even on their phones, while lying on their couch.
With ServiceNow Vancouver being in early availability, we now get a glimpse of the capabilities of ServiceNow’s Generative AI strategy, and what a native LLM allows us to do on the ServiceNow platform.
The main goal of this improvement is to pull AI solutions out of the shadow AI domain, and into a trustworthy and human-centred approach. This allows the use of AI technology while maintaining a secure platform and date, with a centralised governance.
Let’s take a look at some of the ServiceNow Generative AI capabilities.
Case summarisation
A big part of an agent’s job is to deliver accurate and speedy resolutions, but sometimes the “speedy” part suffers from the difficulties of getting a precise overview.
With the help of ServiceNow Generative AI, agents can get a summary of a case or incident, while also receiving recommendations on what to do next, in order to resolve the issue.
This will have a big impact on the manual workload and will boost productivity while accelerating customer outcomes.
Keeping in line with the governance and security concerns of AI solutions, the native Generative AI makes sure that our data is kept safe from external tools.
Now Assist – Intelligent self-help
Many users have already asked existing AI platforms such as ChatGPT for help with writing cover letters, emails, CVs and other documents, utilising the natural language of the technology to speed up the process and give them a head start.
Using this feature in ServiceNow, you can create a Knowledge Base data.
ServiceNow Generative AI will also be expanding upon the functionality of the existing AI-powered Search capabilities. While GPTs can be prone to hallucinations when giving answers, Now Assist for Search provides more accurate natural language responses. This is possible because the answer is based on the customer’s own environment, which greatly reduces the likelihood of errors.
The Generative AI will lower the amount of Agent interactions and allow the customers to help themselves, by quickly finding an answer to a question or assisting with the making or changes to requests.
The Generative AI will be available as an alternative to the Virtual Agent, also allowing internal users to access GPT capabilities directly on the ServiceNow Platform.
These features will help the productivity of internal users while giving external users a satisfying self-help experience.
Text-to-code
Developing code can sometimes be tedious and time-consuming, even when creating the same code for routine scripts and commands. Thankfully, Generative AI is introducing text-to-code, which allows developers to describe what they need the script to do in natural language. This description is later converted into a high-quality code shown directly in the coding frame, ready to be reviewed or further developed by the developer.
The human-in-the-loop approach is still needed, to make sure the code suggestion conforms to the standard practices of your company or customer.
As a part of the text-to-code feature, the developer can also request the Generative AI to complete the currently written code, based on context.
Human in the Loop
As with all AI requests and interactions, whether it is text-to-code, getting help with the Now Assist Agent, or other interactions, we all have to become “prompt engineers” in some capacity. The biggest rule of thumb for dealing with and developing AI is “Good Data -> Good AI, Bad Data -> Bad AI”, which also matters when interacting with Generative AI.
By giving the AI thoroughly descriptive, and sensical prompts, the AI will in turn deliver more correct and accurate suggestions and answers.
To re-emphasise the beginning of this article, the strategy is to have a human-centred approach, to ensure trustworthy solutions through a human failsafe barrier.
Easily configurable
ServiceNow has a focus on low code/no code, and citizen developers. Keeping in line with this strategy, one of the ways to configure the Generative AI is through flows in the Flow Designer.
With flows, a developer can initiate the controller as a simple action, and choose which feature of the Generative AI you wish to invoke.
The types of AI response types seen in the picture are all OOTB.
What you want to do with the output, is completely up to your imagination. Examples, jokes and serious, could be:
- changing the priority of an incident based on the sentiment analysis;
- summarising and giving an actionable list of steps based on the description of a case;
- producing an actionable list of steps to solve an issue before the end user needs the attention of an agent;
- generating a list of jokes and sending it to the requester if the requester seems to be in a bad mood.
The Future of ServiceNow Generative AI
AI technology has been a fundamental part of ServiceNow since 2017, and with the global strides within the field of AI development, ServiceNow is naturally taking advantage of that fact. The Vancouver release introduces a much more obvious and direct link between the user and AI, and more functionalities will surely arrive in our instances with future releases.
So hold on to your hats, as ServiceNow is officially sailing the AI seas aboard SS Generative AI, and look out for upcoming news with Vancouver’s full release.