Case study

Pluralsight's Generative AI Assistant

Leading the design of an AI-powered assistant that delivers personalized content to 7.7M+ users

Role
Senior Product Designer — Generative AI
Timeline
Sep 2023 – Mar 2024
Company
Pluralsight

Context

  • Pluralsight offers 7,800+ courses, labs, and assessments
  • The existing search engine struggled with compound searches, making it hard for users to find relevant content

Problems to solve

Business

Offer AI technology as a competitive advantage in the education industry

Leaders

Curate content for employees that remains current and relevant

Learners

A personalized learning journey based on their current skills and future goals

My contributions

  • Conducted a thorough competitive analysis of established AI tools
  • Communicated identified risks to stakeholders, including data limitations and potential concerns associated with utilizing the Pluralsight search engine
  • Led the team towards a strategic solution, supported by comprehensive research and design principles
  • Made essential adjustments to the feature set in response to deadlines and limitations
  • Refined the design solution through iterative processes, ensuring alignment with user requirements

Competitive analysis

I researched how leading AI assistants approached onboarding, conversational design, error states, and feedback mechanisms to inform our design direction.

Design explorations

I explored multiple approaches for the assistant's interface, testing different interaction patterns to find the right balance between functionality and simplicity.

Final design

The final AI Assistant lives within Pluralsight's navigation, providing personalized course recommendations, answering technology questions, and helping users get started with their learning journey.

Constraints

Courses, labs, skill assessments

LLM hallucinating causing complexity and inaccuracy

Historical context

The LLM retaining historical context was too costly

Saving recommendations

Resource constraints prevented this feature for the MVP

Development resources

Used an existing teams framework knowing we’d get tech debt

Success metrics

Increased user engagement & satisfaction

Uptick in engagement and positive reviews regarding the assistant’s accuracy

AI assistant adoption

Users are 90% more likely to use the AI assistant for content recommendations instead of search

Improved responsiveness

Decreased the average number of messages sent before receiving content back

Enhanced user retention

The average number of sessions per user increased, indicating a rise in returning customers

Recap

What did I accomplish?

I led the development of a successful AI assistant that effectively recommends personalized content.

What skills did I acquire?

How to define success in ambiguity. An understanding of AI and LLM technology. Articulating potential risks and concerns related to the desired business outcome while ensuring successful project delivery.

What was my contribution?

I navigated the business's ambiguous ask of “Build an AI tool” to create a product that solved real user needs.