Build your own natural language emoji search to find the right emojis based on what you mean rather than exact descriptions. Today, most emoji search is based on exact text, not semantic meaning.

For example, if you search for “audio,” no results show up. This isn’t because there are no audio-related emojis. Instead, it's because the specific string of text “audio” does not appear in the name or description associated with any emojis.

We can use multi-modal LLMs to generate richer descriptions and then use Hypermode's built-in embedding and vector search service to build a lightweight natural language emoji search engine.