ServiceNow's AI Chat for Busy Knowledge Workers
Designing a cross-platform, personal AI chat that meets people where they work.
Role: Co-lead designer – Led execution, cross-platform pattern alignment, and strategy.
Timeline: ~6 months
Impact
Unified experience across platforms
Replaced fragmented versions with one scalable pattern, boosting engagement and reducing task-switching fatigue.
Drove business growth
Since Q2 2025: +528 customers (>$5M ACV), +89 >$1M ACV deals, +30% YoY growth in $20M+ accounts
Influenced product roadmap
Design explorations and vision from this work led to new features now planned for release over the next 18 months.
Opportunity
Give Now Assist a home that works everywhere.
A flexible, context aware chat that adapts to needs, surfaces relevant skills, and grows with the platform.
Informative. Proactive. Personal.
Understands context, remembers tasks, and knows preferences. Feeling like part of your workflow, not another tool to open.
Strategy
Designing an AI ecosystem
I co-led the creation of a cross-platform AI engagement model adopted across multiple product lines as the standard for future experiences.
Our guiding pillars:
Unify experiences to eliminate fragmentation
Keep context so AI works within user workflows
Design for flexibility across diverse needs
Future-proof with rich content and proactive assistance from day one.
Established a foundation enabling consistent, scalable AI interactions company-wide.
Process
From alignment to solutions
Getting to the solutions I outline later took cross-org alignment, leadership buy-in, and months of iteration and research. I worked across product lines and platforms, bridging design, engineering, and product teams to define a shared AI model that could scale.
Curious how I worked across multiple teams to drive alignment across platforms?
Let’s chat. I’d love to walk you through it face to face.
Solution 1 of 4
Preserving context
To prevent users from losing their place, we designed a flexible chat window that can switch between views without disrupting the page. It can be expanded, collapsed, pinned, or resized as needed. Letting people work the way they prefer.
Customers can set defaults tailored to their platform and user needs. We audited existing controls and refined them through user testing to land on the final interaction patterns and iconography.
Prototype of window controls for switching between views
Solution 2 of 4
Making entry points intentional
Following our guiding principle of honoring user intent, we stopped forcing users into chat. If they search, they go to search; if they chat, they go to chat.
To meet business goals of bringing the two closer, we surfaced an AI-generated answer at the top of search results, with intuitive entry points to continue the workflow in chat. This gave users a choice, and set us up to recommend choices in the future.
Getting there required extensive iteration and close collaboration with engineering to solve for LLM constraints, enabling inline actions both in chat and on the search page.
Search with AI answer and entry points to chat
Solution 3 of 4
Unifying the experience across platforms
Audited existing designs and partnered with lead designers for portals, workspaces, and mobile to create a consistent, adaptive AI window that worked everywhere without losing platform-specific strengths.
The result:
Workspaces AI was action-focused for faster agent resolutions
Portals focused on conversational answers and forms for quick self-service.
Mobile balanced actions and answers while emphasizing multi-modal input and minimizing distractions.
Now Assist chat is consistent across platforms
Solution 4 of 4
Augmenting chat with rich, context-aware views
Introduced an interactive, expanded view that integrates into existing enterprise workflows. Augmenting chat with immersive, context-aware experiences while keeping users in flow.
To ensure scalability, we created patterns and guidelines for when to use it:
Use when the view adds new context different from the user's current one.
Avoid when it duplicates the current context.
This approach balanced high demand from multiple teams with a consistent, consumer-grade experience across enterprise use cases.
Interactive and expanded view
Bi-directional communication between page and chat
Impact
The Now Assist chat fueled measurable business growth for Agentic AI after its Q2 2025 release.
Enterprise adoption accelerated: The number of customers with over $5M ACV climbed to 528, a ~19.5% year-over-year increase — showing that the unified, scalable Now Assist experience resonated with high-value accounts.
High-value deals increased: In Q2 alone, ServiceNow closed 89 net-new transactions over $1M in ACV, reflecting strong demand for AI capabilities that fit seamlessly into existing workflows.
Strategic expansions grew: Customers with over $20M ACV rose by more than 30% year-over-year, signaling deeper platform investment driven by Now Assist’s ability to deliver value across the enterprise.
This wasn’t just a design upgrade — it was a strategic shift. By making Now Assist consistent, flexible, and context-aware across platforms, we turned it into a product that customers were willing to expand, upgrade, and bet bigger on.
Reflections
Through this project, I practiced…
balancing long-term vision and phased delivery
setting strategy and shaping vision through collaboration and alignment
partnering with research to help quantify the impact of design changes
partnering with PM to prioritize features and create a roadmap
navigating complex, cross-product workflows
