databricks-ai-functions

What problem does it solve?

This Skill removes the friction of adding AI-driven understanding to your Databricks SQL and PySpark pipelines, so you can parse documents, extract fields, and generate business-ready outputs without managing model endpoints.

Core Features & Use Cases

• Document parsing and text extraction: Convert PDFs and office documents into structured page and element content using ai_parse_document.
• Task-specific enrichment: Apply ai_classify, ai_extract, ai_summarize, ai_mask, ai_translate, ai_similarity, ai_fix_grammar, and ai_analyze_sentiment directly in queries for reliable end-to-end enrichment.
• Fallback for complex structured output: Use ai_query when you need nested arrays or complex reasoning that task-specific functions can’t express.
• Production pipeline patterns: Build batch document ingestion pipelines (parse → enrich → extract nested JSON) and custom RAG flows (parse → chunk → index → query).

Quick Start

Ask the AI to extract invoice header fields and line items from a batch of PDF files stored in a Databricks volume, returning clean structured JSON and skipping rows where parsing fails.