As a key member of the design team for ServiceNow GenAI sprint. I contributed significantly to enhancing ITSM workflows with innovative AI solutions. The project is focusing on streamlining workflows and improving problem-solving efficiency through Generative AI. I was responsible to participate in the sprint and help facilitate the workshop, also build a prototype for that and get feedback from the team before moving on to test that with some users.

‍

‍Using Sprints to get through

The team decided to use a design sprint to get results quickly. This sprint was a collaboration between the AI Research team, UX Design and PM teams. We gathered for a week and met every day for around 1 hour. We gathered for a week and met every day for around 1 hour.

Day 1: Align and understand

Initial research

On day 1, the research team partner shared all the relevant insights from their research. My high-level learnings: ‍Support Agents see the time-saving potential of case/incident summarization for quick updates and providing summary notes. Support Agents hesitante in fully trusting summaries, highlighting a need for accuracy to build trust.

‍

Who? Where? What? Why? When? How?

On day one, team aligned on personas(s), current state, problem(s), business context and criteria.

‍

What's stop agents from performing better?

We reviewed all our current personas, discussing our thoughts on each. Following this, we formulated a single, clear problem statement that addresses the needs of multiple personas, as identified from the team's brainstorming.

‍

Day 2 Inspiration and ideas

‍

Time for some sketching

Not everyone feels comfortable sketching. so the team crafted their storyboard with ready to use assets and everyone had their chance to explain their thinking and where they were coming from with it. After everyone went through their ideas, we moved on to voting and I grouped similar ideas under 4 main topics.

‍

Day 3: Voting on what to build

Prioritization

In a follow-up recap, we reviewed the problem statement and ideas from our previous session. It was evident that everyone was well-aligned on the project's direction. By grouping and classifying ideas, we clearly identified duplicates, allowing us to consolidate them into a unified concept.

‍

AI improved expereince could be

As different teams would be impacted by this new feature. I met with cross function team in the wider business to uncover and understand what touch points we needed to consider.

• CASE/INCIDENT HANDOFF: Agent receives case/incident from VA or another agent who needs their expertise or who was unable to resolve

• CASE/INCIDENT RESOLUTION NOTES: Before resolving a case/incident, it is a best practice for agents to provide detailed documentation on the resolution

‍

A potential solution

The big learning! Garbage in equals garbage out! 

A key takeaway from study was the critical importance of input data quality for our AI tools. For AI to assist agents effectively, each incident must be initially summarized with accurate details.

This groundwork enables the AI to provide more precise assistance in return, creating a virtuous cycle of improvement and efficiency. This foundational work is not just about enhancing current processes but is a step towards a smarter, AI-driven operational future.