Skip to main content

Overview

Podium provides multiple search and discovery surfaces powered by PostgreSQL tsvector full-text search with custom ranking. The search system spans products, creators, campaigns, rewards, and collectibles with both traditional keyword search and AI-enhanced universal search.

Query Parameters

When type is omitted, the search returns results across all entity types, grouped by category.

Response

Universal search provides a unified search experience with optional LLM-powered synthesis that generates natural-language answers alongside traditional results.

Query Parameters

LLM Modes

Response with LLM Synthesis

Search Suggestions

Autocomplete

Legacy autocomplete endpoint for backward compatibility:
Prefer /universal-search/suggestions for new integrations.

Discovery Feeds

Curated content surfaces for browsing.

Main Discovery

Category-Specific Discovery

Personalized Feeds

For-You Feed

Returns a personalized product feed based on the user’s intent profile, interactions, and enrichment data. This is the consumer-facing expression of the agentic product feed.

Saved Items (Wishlist)

Users can save items for later across three entity types.

Save an Item

List Saved Items

Remove Saved Item

Supported Item Types

Search Architecture

Search uses PostgreSQL’s built-in full-text search with tsvector columns on SearchDoc and SearchChunk models. Documents are indexed with custom ranking weights for relevance ordering. Products published via the product-published event are automatically indexed. Campaign and creator records are indexed on creation and update. The universal search layer adds optional AI synthesis to generate conversational answers grounded in the search results.

Endpoint Summary