
Leveraging AI to promote self-service across 80+ software products, expanding the Kiwi design system to drive $3M+ in annual support savings.
Bentley Systems provides specialized engineering software used by infrastructure professionals worldwide. With a portfolio spanning 80+ technical products, effective support is critical to their users’ success.
Bentley’s original ServiceNow support portal was disjointed—users struggled with unclear navigation, convoluted ticket forms, and no clear path to urgent phone support. The existing support experience was primarily agent-driven, with scattered self-service capabilities failing to leverage the wealth of knowledge already available.
As Senior Product Designer, I led the end-to-end redesign process with responsibilities spanning:
Product Discovery
Leading the 60+ customer response analysis, stakeholder interviews, and defining pain points.
Prototyping
Building interactive prototypes to demonstrate AI capabilities.
Product Strategy
Working along side Project manager, support director, engineers to envision AI-powered self-service.
End-to-End Oversight
Managing the project from discovery through implementation.
Hi-Fi Design
Crafting iterative and interactive polished visual designs for all support portal components.
Design System Expansion
Extending the design system with support-specific components.
Our analysis identified that 45% of support requests could potentially be resolved through effective self-service, representing an opportunity to save over $3M annually while improving customer satisfaction through faster issue resolution.
80+
Software Products
45%
Potential Self-Service Resolution
$3M+
Potential Annual Savings
AI-Powered Self-Service
Implement intelligent search and recommendation features to help users find solutions without agent intervention
Cost reduction
Reduce support costs by $3M+ annually by enabling effective self-service resolution
Unified Experience
Create a cohesive support experience across all 80+ software products while respecting product-specific needs
Design system evolution:
Expand the Kiwi design system to include support-specific patterns and components
To deliver an exceptional support portal experience, we must deeply understand our users’ needs and pain points. This phase ensures we’re aligned with internal stakeholders and informed by user insights.
Shown below is the entire service flow blueprint illustrating the complex journey users navigated. My work focused on simplifying entry points and critical user paths while respecting the existing backend technical resolution processes.
Simplified user flow from various touch-points.
As part of my discovery process, I conducted a comprehensive UX audit of the existing ServiceNow portal to identify specific pain points and improvement opportunities. The audit revealed several critical issues affecting user experience:
To guide our design decisions, two distinct user personas we developed based on research: a manager overseeing infrastructure projects and a power user working directly with the software daily. For this case study, I’m focusing on the manager persona, as they represent a critical decision-maker who needs efficient support solutions.
For the ideation stage, I worked with stakeholders to brainstorm and ideate ideas together. The process was made super productive by engaging developers at the beginning because design solutions for quick wins needed to be built around ServiceNow constraints. Together, we developed a two-phase implementation strategy to deliver immediate value while building toward our long-term vision.
Phase 1: Quick Win Implementation
To deliver value quickly while working toward the full AI vision, team focused on cohesive experience improvements with layout and branding enhancements that were delivered in November 2024.
Personalized Software View
Implemented a personalized dashboard showing only the software products the user owns, dramatically simplifying the support experience for users of Bentley’s 80+ software offerings.
Cohesive Branding
Applied consistent Kiwi design system patterns to the ServiceNow portal, creating a cohesive experience while adding new support-specific components to the design system.
Prominent Emergency Support
Redesigned the portal to make phone support for urgent needs clearly visible, ensuring users can quickly find help during critical outages or time-sensitive issues.
Mockup
Phase 2: AI-Powered Self-Service Implementation
Leveraging AI to Promote Self-Service
With quick wins delivered. I focused shifted towards is transformative AI implementation to promote self-service options and achieve the projected $3M+ annual cost savings
Conversational Interface
Designing an AI chat experience where users describe their issue in natural language and receive targeted solutions
Designing seamless transitions to live agent interfaces when AI self-service can’t resolve the issue, preserving all context
Creating systems that automatically provide video tutorials, articles, and community knowledge base content based on the user’s issue
Building AI systems that understand relationships between Bentley’s 80+ software products for comprehensive support
Interactive Prototype
Key Design Innovation
Personalized AI Experience
Designing an AI interface that adapts to user’s software portfolio, expertise level, and usage patterns to provide highly relevant support
Critical Path Escalation
Designing clear pathways to phone support for urgent issues while encouraging self-service for routine questions
Multi-Modal Support Resources
Creating a system that intelligently delivers the right format (video tutorials, documentation, community knowledge) based on the issue and user preferences
Business Impact
Project Manager, WSP
Our redesign approach addressed Bentley’s support challenges through a two-phase strategy. Phase 1 delivered immediate improvements to the ServiceNow portal in November 2024, including personalized software dashboards, cohesive branding with the Kiwi design system, and prominent emergency support access. Phase 2 (currently in development) focuses on AI-powered self-service through a conversational interface that intelligently delivers relevant support resources while maintaining seamless escalation to live agents when needed.
Lift:
22% increase in self-service utilization after Phase 1
32% decrease in navigation time
$3M+ projected annual savings upon full implementation