Custom AI Extraction — Deploy the GPT-4.1 Function — Build an AI Product Search Agent (Part 5/6)

This is the core of the project. We deploy the Azure Function to Azure, test it with a sample payload, then update the skillset with a Custom WebApiSkill that calls GPT-4.1 for structured extraction. We add 37 output field mappings to the indexer, re-run it, and watch every product field go from NULL to fully populated — SKU, color, VOC, dry times, warranty, and more.

What you'll learn:

Deploying an Azure Function with func azure functionapp publish
Testing the custom skill with a sample payload
Adding a WebApiSkill to the skillset (GPT-4.1 extraction)
Configuring 37 output field mappings in the indexer
Re-running the indexer and verifying all fields populated
Before/after: NULL fields → structured product data
🔗 GitHub Repo: https://github.com/dhavalshah01/contoso-ai-paints

Видео Custom AI Extraction — Deploy the GPT-4.1 Function — Build an AI Product Search Agent (Part 5/6) канала Dhaval Shah