Have you heard? Generative AI (GenAI) has been on the tip of everyone’s tongue lately! Many folks have strong feelings about GenAI and AI, in general: love, hate, fear, just to name a few.
On August 15th, 2023, the ServiceNow Dallas Office hosted an interactive workshop to “discuss how to harness the power of Generative AI for good low code app development.” We heard from several thought leaders, including ServiceNow’s Omar Martinez and Jeff Margolese, joined by a few Cask employees.
Is it possible to use GenAI for good (and responsibly)? Yes, of course, but there are no governing laws or even a world consensus on how “good” and “responsible” are defined. I recently watched a fascinating discussion, including ServiceNow’s Chief Innovation Officer, Dave Wright, when he sat down with Dr. Catriona Wallace, a highly respected AI and Machine Learning entrepreneur. I highly recommend you watch the recording of this discussion and read Dave’s article on the topic; both can be found here: A digital gold rush will be built on AI.
AI Uses For The Enterprise
(GenAI) is artificial intelligence capable of generating text, images, or other media, using machine learning. Generative AI models learn the patterns and structure of their input training data and then generate new data with similar characteristics. Read more on Wikipedia…
In Dave Wright’s article, he suggests that the best enterprise use case is using GenAI to reduce the time spent on routine administrative tasks. Things like summarizing an ongoing issue, pouring through a lengthy employee handbook to find a policy for your specific location, or gathering details from receipts to generate an expense report automatically. Typically employees don’t relish doing these time-consuming tasks, and this time wasted could be spent productively.
During the workshop, we discussed how ServiceNow features would make automating these types of tasks a reality!
- Case summarization: Who wants to analyze a ticket’s history to create a paragraph write-up? Not me!
- Conversational exchanges: Can you imagine asking a Virtual Agent (VA) about a specific detail of an HR policy? Typically, it might provide a link to the handbook in full or copy the text of the entire policy into the chat for you to read and decipher. With GenAI, the VA can understand who and where you are and provide you with only the relevant information.
- Content creation: Have you ever seen the snippet at the top of some Google searches? AI analyzes details from top websites for that search and creates a short rundown of the highlights. ServiceNow can also do this with your company’s Knowledge Base Articles.
- Text-to-code: generate code from a natural language statement like “find all of the Priority 1 Incidents opened in the last 30 days” Though I consider myself very proficient in ServiceNow code, things like Date/Time logic always trip me up, and this would be a great time saver.
- Generative AI Controller: If the above features that ServiceNow offers aren’t quite what you need, you can use the GenAI controller to connect to OpenAI, Azure OpenAI, or the ServiceNow large language models. This will be available within Flow Designer, Virtual Agent, and scripts to tailor a solution to your specific requirements. The controller can be used to perform a number of functions including Create Content, Generic Prompt, Summarize, QnA, and Sentiment Analysis.
I am eager to see how companies creatively use these features to improve their processes. We divided into teams in the workshop to develop our ideas and then presented them to the group. Here are the ideas the teams came up with:
- LegalEats AI: an app to help restaurant workers navigate and understand legal actions and documents regarding on-the-job incidents and injuries.
- The Sharpies: An Employee Onboarding app that eliminates duplicate tasks, provides a helpful virtual agent, and ensures a smooth start to each employee’s journey.
- ElevateXperience: This app incentivizes remote workers and travelers to return to the office, and therefore making better use of real estate that the company is already paying for! The app would provide features like: way-finding to the nearest printer, local recommendations for dining and activities, networking with fellow hot deskers, and easy access to support across the enterprise.
- Only Fish: A fishing club app that promises a premium fishing experience. Letting club members know where the fish are biting each day and allow users to reserve time slots so that everyone gets a fair chance to hook a fish.
How do I start?
What a variety of ideas our teams came up with! Imagining completely new concepts to use AI is fun, but you don’t have to start a massive new project to start using AI within your instance. ServiceNow has been acquiring AI-related technologies since 2017. Chances are you already have access to AI features you aren’t using today!
To begin your ServiceNow AI journey with robust features that are already available, Jeff Margolese recommended starting with the following steps:
- Implement AI Search – This feature is available with some configuration in Service Portal, Now Mobile App, and Virtual Agent. Also, developers can add AI Search to UI Builder pages and extend AI Search with ServiceNow Store applications and integrations.
- Review and curate your Knowledge Management Strategy – AI solutions need large amounts of data to learn from, hence the term Large in Large Language Model (LLM). The outputs of your AI models will only be as good as the data you feed into them, so you may have some prep work to do before flipping the magic AI switch.
- Use Configurable Workspaces – Though some areas of the “classic” Platform interface will have access to future AI features, ServiceNow’s continued development investment is going into Workspaces. “Solve issues faster with a workspace built to facilitate resolutions. From a single view, get full context, analytics, and AI-assisted recommendations.”
- Deploy Virtual Agent – AI already powers Virtual Agent (VA), and in a future release, it will be enhanced by GenAI to make the conversations with the chatbot even more helpful and personalized. Not to mention adding VA to your self-service arsenal typically reduces Service Desk calls, resulting in increased productivity for all!
- Consider other AI-powered solutions – Robotic Process Automation (RPA), Document Intelligence, and Sentiment. You’ll need to evaluate whether any of these solutions are a good fit for your specific environment.
Check with your ServiceNow Sales Rep to understand what AI features you can access with your current subscription. I know ServiceNow is eager for its customers to supercharge their workflows with these solutions!
Data Security
At Serenity, we are also using the power of GenAI to boost our customer’s success. At ServiceNow Knowledge ’23, we debuted two new use cases in our suite of applications: AI-driven inspection templates and smart corrective action recommendations. I am thrilled to be able to make these features available to our customers to increase their adoption and efficiency on the platform. This solution will be able to connect to an LLM of the customer’s choosing.
At this point, we’ve all heard about at least one well-intentioned employee that freely copied private and confidential company information into ChatGPT. In case you hadn’t heard anywhere else, the data you enter into ChatGPT (and other public GenAIs) IS NOT KEPT PRIVATE! As a result, many companies have been considering private Large Language Models (LLM) to protect their sensitive data and preserve their competitive advantages.
ServiceNow knows this will be a concern and has partnered with NVIDIA to develop a custom LLM that will be available to its customers. ServiceNow’s AI solutions will depend primarily on this model, and this will be a boon for customers because they can access a much larger data pool created from the data of many ServiceNow customer instances. Remember, with LLMs and GenAI, the larger the data set, the better the results. ServiceNow’s LLM will reside in the same data centers where their instances are already hosted, meaning queries will not be sent to public LLMs like the one ChatGPT uses.
Details around how this works are not evident yet; we did hear there will be some flex capability to use OpenAI or Azure OpenAI as an alternate LLM for some features. Of course, customers can also use the Generative AI Controller to connect to any LLM.
There are still a few security concerns to address before I think any company will jump at the chance to adopt any LLM. Do customers know their data is being used for ServiceNow’s LLM? Typically code doesn’t contain any private or confidential information, but I’m not sure we can say the same for IT Incidents and CSM Cases. No one will be able to access data in the ServiceNow LLM directly, only use it through prebuilt features or the Generative AI Controller. According to ServiceNow, only about 2% of customers have opted out of sharing their instance data for the LLM.
What’s next?
If you are ready to get working on AI in your instance, ServiceNow also has a wealth of resources in the Enable AI documentation landing page and the AI & Intelligence section of the Community. From a simple to navigate “Getting Started” landing page to links about Advanced Topics and Academy recordings.
Additionally, ServiceNow has training available on NowLearning for many of their AI solutions, including two beginner courses:
Though there are some unknowns and concerns around AI in general, I still recommend staying on top of the details and familiarizing yourself with the technology. You might be left behind if you use the “I’ll just wait until it’s all figured out” approach! At the very least, you will be able to speak more confidently on the topic, know the capabilities, and be ready when your leadership comes knocking.
