Agentic AI implementation Assessment

[GokulNathV]

Hi all

We are about to implement agentic AI in ITSM for our client. Before implementing what are the assessment/questions to be asked to client? Thanks in advance

[Clifford Tiang]

Hi @GokulNathV ,

ServiceNow has some great Now Assist best practices and deployment guidelines found here - https://mynow.servicenow.com/now/best-practices/collections/now-assist-best-practices

Just to share some personal 2 cents tho. In the time of Agentic AI, everything becomes outcomes driven.

I will want to understand the outcomes that the customer is trying to achieve and move forward from there, because the outcomes will drive the kind of usecase that should be priortized. 

Could it be to reduce triaging efforts? or to reduce MTTR? or it could be, to make sure that all human agents are putting the right info into the tickets to beef up the data. There's many ways that this conversation can go, and many corresponding ways to solution.

Bottom line is, i think we need to be very deliberate with Agentic AI usecases so that we can really help customers derive value out of it and drive adoption. Also, do make sure to take snapshots of metrics pre and post deployment so that these value metrics can be easily quantified and showcased back.

All the best!

[GokulNathV]

Hello Clifford

Thanks a lot for the share and the tip.
Unfortunately we are yet to touch base with our client and in progress for due diligence phase. But will sure get back once i have more context.

Thanks once again!

[Tanushree Maiti]

Hi @GokulNathV 

I guess following questions could be asked. 

Use Case Definition

• What outcomes are expected from each AI agent?
• Which types of human errors can AI help reduce or eliminate?

Data Foundation

• How dependable and well-structured is the current knowledge base for training and decision-making?
• Which third-party systems or APIs need to be integrated, and how will secure access be managed?

Governance & Oversight

• What governance framework will be used to audit and review AI-driven decisions?
• Who will be responsible for monitoring, validating, and approving agent actions?

Technical Readiness

• What approach will be used to test and validate agentic workflows?
• Will AI agent evaluation frameworks, simulations, or automated testing mechanisms be leveraged beyond manual testing?

Organizational Value Addition

• How will the success of AI agents be measured?
• Which KPIs will be tracked, such as MTTR reduction, automation coverage, operational efficiency, or user satisfaction?