All notable changes to this project will be documented in this file.
- Added: Option to install just some specific GPT-RAG components: data ingestion, orchestrator and frontend.
- Added: Optional Reuse of Pre-Created Resources: You can now optionally reuse existing resources such as VNets, Azure OpenAI instances, etc.
- Added: Use five subnets to align with the idealized architecture.
- Added: Custom Addressing to allow specifying custom addressing for VNets and subnets, providing greater flexibility in network configuration.
- Added: AI Integration Hub to leverage the power of various external data sources to enhance its capabilities. Currently, we have integrated the following products:
- AI Search: Enables access to vast online databases, providing precise and up-to-date information.
- Bing Custom Services: Customizes search results for specific business needs, ensuring relevance and quality.
- SQL Service: Queries extensive internal databases for accurate and current organizational data, performing analytical functions like count, sum, average, and more.
- Teradata: Integrates large-scale data warehousing capabilities, enhancing data retrieval and analysis.
- Added: AOAI content filtering and blocklist configuration.
- Changed: Changed sample documents and test dataset.
- Added: Load testing.
- Changed: Increased Zero Trust VM SKU to 4vCPU and 16GB.
- Added: Blob storage data source soft deletion support.
- Changed: Disabled blob anonymous access.
- Changed: AI Search API version defaults to 2023-10-01-Preview for indexProjections and MIS authResourceId support.