Building out a new Autocomplete & Search / ‘Jump-To’ experience for Slack was a fun project! Performance and offline support were top of mind.

In case you’re curious, autocomplete involves tagging people, channels, etc.

To prepare Slack for lazy-loading data like users and channels, we couldn’t rely on synchronous assumptions.

The ‘data providers’ stream all sorts of models, composing the object graph internally so that feature teams don’t have to!

Components ranging from something as small as a button, to a performant declarative, data-driven list.

A ‘Playground’ executable is bundled for internal eyes only, allowing engineers, product, and design to explore the library of components.