Enhance the Spotfire Copilot PDF Data Loader to support configurable OpenAI-compatible API endpoints, including internally hosted LLM gateways such as LiteLLM.
Currently, the Data Loader’s OpenAI plugin does not support custom base URLs, limiting integration with enterprise-hosted embedding models.
Our environment uses an internal LiteLLM gateway that provides embedding models through an OpenAI-compatible API (/v1/embeddings) with API-key authentication.
However, the PDF Data Loader cannot be configured to use this endpoint. This prevents us from utilizing our existing embedding infrastructure for PDF ingestion and retrieval.
The suggested Ollama alternative is not suitable for our environment due to internal platform restrictions.
Allow administrators to configure:
A custom OpenAI-compatible API base URL.
API-key or bearer-token authentication.
The embedding model name or identifier.
Offline tokenizer support without requiring external internet access.