Ally Zhu

Summary

Search - finding what you need when your team is scaling and context gets lost - is overwhelmingly why team customers pay for Playbook. It’s also our hardest and most ambiguous problem. Search queries come in all shapes and sizes, with some better served by direct filtering and keyword search, and others by natural language agentic search.

To serve those disparate needs, we historically had 3 entry points for search: filtering, search bar, and agent. This was causing paralysis – users had to pick a search experience before they knew which one their query needed. For queries that went through, we saw frustration from unpredictable results: users were searching by natural language what were best served by filters.

It was time to rethink our search experience: could we create one surface that reduced paralysis, retained simplicity, but increased customizability? How could we better translate user intent into the most suited search experience?

I answered those questions by redesigning search. Type “all PNGs uploaded by Ally” and Playbook predicts the filters you mean and turns them on as you type. Describe what you see and AI results fill the same page. Ask a real question and you get an answer, with the option to keep talking to Playbook.

Context

Finding what you need when your team is scaling and context is lost is the reason team customers pay for Playbook. It is also the hardest and most ambiguous problem in the product. The more assets a team stores, the more they depend on search, so search quality decides whether usage holds as a library grows.

By mid-2026 we had three ways to find things, each built at a different time for a…

This post is protected.