Build on the discourse graph

Submit a natural-language query and receive the most semantically relevant discourse records, ranked by vector similarity. The query is embedded in real time using Voyage AI, then matched against the full corpus via pgvector. Results include relevance scores, source metadata, and timing diagnostics. Use this when keyword search falls short — for exploratory research, concept mapping, or finding related conversations across platforms.

Returns a paginated feed of all published narratives across every content type AIDRAN produces. Filter by narrative type, editorial beat, date window, or featured status to surface the stories that matter most — whether you are building a dashboard, populating a feed, or auditing editorial output over time.

Returns a paginated feed of intelligence signals detected across all editorial beats. Filter by topic, severity level, or detection date to zero in on the moments that matter — a sudden spike in safety discourse, a sentiment reversal around a product launch, or an emerging narrative that has not yet been covered. Each signal includes its type, severity, a human-readable headline, and structured detail.

Returns the most-discussed entities across AIDRAN's discourse corpus, ranked by mention count and filterable by topic and time window. Each entry includes the entity name, mention total, and type classifications. Use this to build leaderboards, spot rising players in a beat, or compare mindshare across organizations and technologies.

A lightweight endpoint that returns daily discourse volume for a given topic — just date-count pairs, plus any external trend signals detected in that window. If you only need a volume sparkline or a quick activity overview, this is faster and simpler than /snapshots. For richer data including sentiment and entity breakdowns, use the snapshots endpoint instead.

A comprehensive portrait of a single editorial beat. Returns the current beat narrative, recent intelligence signals, a 30-day sentiment trajectory, the most-mentioned entities, volume broken out by source platform, and active narrative clusters. This is the richest single-call view of any topic in the API — ideal for building beat pages or research dashboards.