Features
Native vector search for AI applications
Full-text search with relevance ranking
Hybrid search combining vector and lexical search
Native reranking for improved retrieval accuracy
Multi-cloud database deployment (AWS, Azure, Google Cloud)
Stream processing for real-time data pipelines
Data federation across multiple sources
Built-in charts and visualization
Document model with flexible schema
Horizontal scaling and load balancing
Enterprise-grade security and compliance
Atlas SQL Interface for SQL-based querying
Online Archive for cost-effective data tiering
Relational Migrator for database migrations
Search and vector search available on-prem (Enterprise Advanced, Community Edition)
Multimodal search across video, images, audio, documents
Auto-extract scenes, faces, OCR, transcripts, embeddings
Multi-stage retrieval: filter, join, rerank in <100ms
Managed Vector Store (MVS) with dense, sparse, BM25 search
Multimodal Extractor v2 with Gemini Embedding 2 (3072D embeddings)
Face & person search across video libraries
IP & copyright detection for logos, songs, faces
Brand & ad safety pre-publish screening
Speaker diarization aligned to transcript timeline
Layout extraction from PDFs (header, body, charts, signature)
Plugin Marketplace for custom extractors
Cluster visualization with color-by-field, similarity slider
Cross-modal joins at same object and timestamp
Source adapters: Iconik, Webhook, Email, Supabase
Deploy-resilient batch jobs with auto-resubmit
