Mixpeek provides infrastructure that allows developers to construct their own search experiences, aligned to their business. It enables this via three layers:

  1. Feature Extraction - Extract meaningful information from any content type using state-of-the-art models (OSS, BYO, or Mixpeek proprietary)
  2. Entity Enrichment - Enhance extracted features with domain knowledge through manual classification, automatic clustering, and relationship mapping
  3. Search & Discovery - Build powerful search experiences with vector similarity, filtering, grouping, and advanced capabilities like Learning to Rank and Fine-Tuning.

1. Multimodal Feature Extraction

Features are the core primitive in Mixpeek. They represent extracted information from any content type using state-of-the-art AI models.

Example: Video Processing

Learn more about Feature Extraction →

2. Feature Enrichment

Once features are extracted, they can be enriched with domain knowledge through three main mechanisms:

Taxonomies

Manual classification using predefined hierarchies for strict content organization

Learn More →

Clusters

Automatic content grouping using similarity and patterns

Learn More →

Relationships

Define connections between features, nodes, and clusters

Learn More →

3. Search & Discovery

Enriched features enable powerful search capabilities:

Search Configuration Example

Learn about Filtering →

Learn about Grouping →

Best Practices

1

Plan Your Feature Extraction

Choose appropriate models and intervals based on your content type and needs

2

Design Your Domain Model

Create taxonomies and relationships that reflect your business logic

3

Optimize Search

Use filters, grouping, and reranking to improve result relevance

4

Monitor & Iterate

Track search interactions and refine your implementation

Need help getting started? Check out our Quick Start Guide or contact our support team.