GROWTH / GORGIAS • 2024
AI-generated Article Library
This project highlights leading user testing and collaborating with Growth to drive feature engagement and visibility.
ROLE
Product Designer
TIMELINE
January - March 2024
COLLABORATORS
Lenaig L., Product Manager
3 Engineers
Machine Learning Team
Growth Team
If you've ever shopped online, you've probably relied on a Help Center at some point to find answers about store policies or product details.
Behind the scenes, a lone support agent or lean team spends countless hours creating, updating, and organizing all that content.
Those articles make self-service possible—but they're also the biggest barrier to building and managing a Help Center in the first place.
They don't know where to start
Support tickets and analytics reveal what customers actually need help with, but merchants lack the people resources to track patterns and translate them into articles.
THE SOLUTION
An AI-generated article library built from real support conversations
Articles are generated by analyzing a merchant's support history—identifying frequent questions and drafting content based on how their agents actually respond.
New AI Library tab
A new space in Help Center settings where merchants can access all their AI-generated articles. No more hunting through half-baked templates or starting from scratch.
View & edit the most impactful topics
Agents can browse the full list of generated articles (highest tickets first) and edit them directly in the interface, turning hours of writing into minutes of refinement.
Quick actions
Archive topics that aren't suited for self-service and publish the rest with a single click.
Smarter onboarding
Instead of generic templates, the onboarding flow now suggests merchants' top 5 AI articles.
Proactive nudges outside settings
Since merchants infrequently manage their Help Center, we surface new AI articles in high-traffic areas like their analytics, homepage, automation dashboard, and via email.
RESULTS & IMPACT
The library was fully released in April 2024.
AI articles became a key step for AI Agent knowledge and optimization, while also enabling customer success teams to onboard customers faster.
5% increase
in Automate merchants with ≥ 10 articles
Building on the ML system
Post-launch, we extended the ML functionality to generate help articles into new features and surfaces of the product, expanding the functionality without any additional model work to push for adoption.
LEARNINGS
When in doubt, talk to users
This project had many unknowns, and competitor insights offered little clarity.
When stakeholder opinions conflicted, user testing cut through the noise, providing the evidence and confidence needed to align decisions and move forward with leadership support.
Design for the full journey
Rather than hoping users would discover the new feature in settings, we considered the broader user journey.
Partnering with the Scaled team helped us create intentional entry points that drove discovery and engagement in relevant places agents frequently visit.