Conversational & Agentic AI
Product Design
Designing AI-native interfaces at enterprise scale — from quick answers to complex, multi-step agentic workflows
TriNet helps businesses support their people and operations.
Trusted by 23,000+ businesses, supporting 500,000 people—helping power the operational heart of the U.S. economy.
Total Customers
23,000
Daily Active Users
3,500
Employees in Platform
460,000
Annual Payroll Processed
$72B
Agentic and AI Experiences
Building security, accuracy and compliance into AI products
I led end-to-end design for TriNet's AI experiences, turning complex problems into intuitive products for 500k+ users where limited conventions existed.
Context
TriNet had three separate AI entry points — search, dashboard, and support — each with different interfaces and no shared context. Users lost context, didn't know which tool to use, and fell back to support instead of getting answers in-product.
Role
Partnered with product leadership and owned the design and interaction exploration — translating the strategy into actual surfaces users could navigate, establishing new interaction patterns for 500k+ users where no conventions existed.
Approach & Outcomes
Early 2026 AI dashboard beta resulted in an 11% YoY decrease in case volume. AI chat experiences project a 50% automation rate, a deflection of up to 29,000 chats annually – allowing users to complete tasks natively without losing context.
Quick Answers
A dashboard focused on surfacing quick answers, reducing the need to hunt through menus.
Voice & Chat
Allowing users to switch between text and voice to complete multi-step tasks, like requesting time off.
Assistant (Bottom Bar Exploration)
A persistent helper that can handle complex tasks in the background.
Assistant (FAB Exploration)
A persistent helper that can handle complex tasks in the background.
Other scenarios and edge cases
Successful quick answer
Unsuccessful: No info available + feedback
Unsuccessful: Do not understand
Dark mode and other interactions
Autosuggestions
Animated loaders
Other relevant links and feedback
Dynamic Dashboard
Interaction design for a conversational search, with dynamic dashboard
Inline Context
An exploration of page-aware AI. The system can read the user's specific time off plans to provide tailored feedback on compliance or other topics.
Thanks for viewing!
iszimberg@gmail.com
