LAB1726: AI All Day - Exploring Prompt Engineering

[AJ Siegel]

What a great knowledge! If you were at our “AI All Day” session you got to customize NowAssists prompts in the platform before it was released to the general public. It was a very productive session. We had just shy of 100 attendees at the lab. They had multiple roles in their org from Developer to Admin and even a couple of people from the C-suite! Their level of experience with AI ranged from very little to a few people who could teach us a thing or two.

AI and its impact:

• We started out talking about AI and prompt engineering basics. From there we looked at not only the value of AI to your organization, but how you can use it to improve your skillset and career.
• We had a very robust discussion about ServiceNow’s history in the AI space (we started heavily investing in 2017) as well as some of our future vision. Specifically, we looked at the shift from preconfigured to configurability in NowAssist which prepared us quite well for our workshop.

Configuring prompts to operate at scale:

• In the lab we used a custom ServiceNow instance developed specifically for this class.
• We identified datasets that were specific to our use case and set the parameters for our AI instance.
• We iterated on a prompt so that it pulled both structured and unstructured data to give us the most accurate answers possible.
• We defined our output and set context to create a richer end user experience.

Evaluation to ensure the right experience and streamline deployment.

• We learned about multiple prompt evaluation methods and evaluated the prompts we created.
• We produced reports on those evaluations and talked about how we create governance to ensure prompting at scale.

Resources

• AI Skills Forecast
• ServiceNow GenAI
• ServiceNow GenAI Community
• ServiceNow AI Documentation

We would love to hear from you if you were there or if you have any questions. The feedback so far has been great.