Knowledge
Knowledge Integrations
Connect with sources of KNowledge and unstrcutured data through your data warehouse, file uploads and native API integrations.
Warehouse Integrations
Warehouse integrations provide a powerful way to process and analyze unstructured data stored in your data warehouse:
- Native Source Support: Process data from various sources (Slack, Notion, etc.) directly from your warehouse
- Source-Specific Processing: Each source type (e.g., Slack messages, Notion pages) has its own optimized processing pipeline
- Metadata Preservation: Maintain source-specific metadata and context throughout the processing pipeline
- dbt Package Support: Leverage dbt packages to transform and prepare unstructured data for processing
API Integrations
API integrations enable real-time processing of unstructured data from various sources:
- Direct Source Connections: Connect to sources like Slack, Notion, and other platforms via their APIs
- Real-time Processing: Process and analyze content as it’s created or updated
- Source-Specific Chunking: Apply optimized chunking strategies based on the source type
- Metadata Enrichment: Automatically enrich content with source-specific metadata
File Uploads
File uploads provide a flexible way to process individual documents and content:
- Direct Processing: Upload and process individual files or batches of files
- Document Intelligence: Extract insights from various document types (PDFs, Word docs, etc.)
Artifact-Specific Processing
Push.ai applies artifact-specific processing and chunking techniques:
- Message Processing: Process chat messages with channel-level artifact grouping
- Document Processing: Handle documents as single artifacts with recursive LLM-based chunking
- Source-Agnostic Analysis: Apply consistent analysis techniques across different sources
- Context and Metadata Preservation: Maintain relationships and metadata across artifacts and source context