Dynamic Home Pages

Ideas don’t die because they’re bad — they die because there’s no time or tooling to prove them.
This project showed what happens when AI closes that gap.

Challenge

For years, designers across the Platform and solution teams had shared concepts for making Kibana’s home page more dynamic and relevant. The vision was to surface user-specific content such as “recent dashboards,” “saved searches,” or contextual entry points based on the data they worked with most.

Everyone agreed it was a good idea. It just never made it to the top of the roadmap. Without engineering time to prototype it, the concept sat untouched in Figma files, GitHub issues, and old Slack threads.

Idea

Revisit the original ideas and rebuild them as working, in-product proofs of concept using AI-assisted coding.

My goal was to show that personalized home pages weren’t just design aspirations — they could be real, functional, and achievable within the existing Kibana architecture.

I used ChatGPT to help sketch the architecture and Cursor to scaffold live prototypes in a local Kibana instance. Within a few short demos, I had fully functional examples showing dynamic content based on user context — all running in code, not mockups.

AI Workflow

• ChatGPT helped identify where and how the home page components were rendered, how content modules were fetched, and where new data hooks could integrate safely.
• Cursor handled scaffolding the POC components, wiring into existing services, and debugging rendering and routing issues.
• I iterated directly inside a live Kibana instance, creating multiple prototype versions to demonstrate different personalization strategies.
• AI tools made it possible to move from idea to working demo in days — something that had sat dormant for years.
• Presented the prototypes to the entire Platform team, sparking discussion around new content management capabilities and where they should live.

Outcome

• Functional POC for dynamic home pages expanded to include Discover, Dashboard, and Kibana Home
• Showed real-time examples of personalized, user-aware layouts in code
• Demonstrated the feasibility (and current limits) of Kibana’s content management approach
• Catalyzed cross-team alignment on next steps for user-specific content architecture
• Proposed new, portable UIs to front existing servies for favorites and recents
• Revealed a hidden architectural issue: the TableListView component used on several pages was an entire page implementation, not a reusable component — blocking adoption elsewhere
• As a result, new planned work is now underway to refactor TableListView into modular, reusable pieces while engineers have joined the exploration of new content panels

Reflection

This project captured what I love about designing in the product.

AI tools made it possible to un-shelve design ideas spanning several years and prove them in real, working code. Seeing the team’s reaction — and watching those prototypes evolve into planned engineering work — was deeply rewarding. Sometimes, all an idea needs is someone willing to bring it back to life with the right tools.