ServiceNow / Now Assist Context Menu

From a writing request
to a shared AI capability.

I saw a platform opportunity inside a request to regenerate resolution notes. I initiated and led the design of a reusable AI component that helps people work with content without leaving their task.

My role: Lead product designer
My contribution: Concept, direction, and detailed design
Timeline: About one year from concept to general availability

Generated content appears beside the original description, ready for review before replacement. — Assistance in context, with an explicit choice before changing the record.

Reframing the request

The request was a trigger.
The opportunity was bigger.

Resolution notes were generated automatically, but the underlying records did not always contain enough useful information. The initial proposal was to let agents choose when to generate them.

I saw a broader possibility: assistance that could begin with a person’s intent, refine something they had already written, or generate content when the context was ready.

The initial request

Let an agent trigger resolution notes.

A focused solution to a problem in one workflow.

The direction I proposed

Make contextual assistance reusable.

Support different intents and content types across the platform.

Making the idea tangible

Give other teams
something to respond to.

I created a lightweight concept and shared it early with designers across the platform and business units.

Their use cases helped demonstrate that the same interaction could support resolution notes, chat, email, knowledge articles, analytics, and more. Those conversations helped build support for a shared component.

Resolution notesChatEmailKnowledge articlesAnalytics

03 / Designing the experience

Keep assistance close.
Keep people in control.

Keep the person in their workflow.

I kept assistance close to the field so people could work with AI in the context of the task.

Separate generation from commitment.

People could review and refine a result before inserting or replacing content. Generating a response did not have to mean accepting it.

Support different starting points.

The experience accommodated both generating new content and refining existing text, giving adopting teams a shared interaction to build on.

A generated comment can be shortened, elaborated, or adjusted in tone before insertion. — Review. Refine. Then choose whether to insert.

Carrying it through

Make the shared idea
work in real conditions.

I partnered closely with Product and Engineering on behavior, configuration, technical constraints, build reviews, and scope.

Scenario-based guidance helped teams adapt the component to different models, prompts, and use cases. That meant thinking through the less obvious details as well as the main interaction.

The outcome

A shared foundation
for AI in everyday work.

The component reached general availability in about a year and was adopted across business units.

Reuse across products

Teams could build on a shared capability instead of designing the same interactions independently.

Room to grow

The foundation expanded from writing assistance into contextual AI across enterprise workflows.

Strategy carried into delivery

The broader platform direction stayed connected to the behaviors, configuration, and interaction details needed to ship.

Continue exploring

How people supervise AI at work.

Explore agentic AI