Simplify API key requirements for students - avoid requiring both OpenAI and Google API keys

[jxnl]

Problem

Currently, students need to set up both a GOOGLE_API_KEY and an OPENAI_API_KEY to run the topic modeling process:

• GOOGLE_API_KEY for Gemini 2.0 Flash (used by Kura for summarization)
• OPENAI_API_KEY for Text-Embedding-3-Small (used for clustering) and classification work

This creates unnecessary friction for students who may:

1. Not have access to both API providers
2. Face budget constraints with multiple paid services
3. Experience setup complexity that detracts from the learning objectives

Proposed Solution

Option 1: Standardize on OpenAI (Recommended)

• Use OpenAI's models for both embeddings and summarization
• Replace Gemini 2.0 Flash with GPT-4o-mini or GPT-4o for summarization
• Keep OpenAI Text-Embedding-3-Small for clustering
• Only require OPENAI_API_KEY

Option 2: Standardize on Google

• Use Google's text embeddings for clustering
• Keep Gemini 2.0 Flash for summarization
• Only require GOOGLE_API_KEY

Option 3: Provide fallback options

• Allow either API key to be provided
• Automatically select appropriate models based on available keys
• Provide clear documentation on which option to choose

Impact

This change would:

• ✅ Reduce setup friction for students
• ✅ Lower cost barriers (students only need one API provider)
• ✅ Simplify the getting started experience
• ✅ Maintain the same learning outcomes

Current Usage

From the README:

You will need a GOOGLE_API_KEY or an OPENAI_API_KEY for running this topic modelling process. We're using the OpenAI Text-Embedding-3-Small embeddings for clustering and the Gemini-2.0-flash models for summarisation (used by kura).

The documentation says "or" but the implementation actually requires both keys for the full workflow.

Files to Update

• README.md - Update installation instructions
• Notebooks 1, 2, 3 - Update prerequisite sections
• Code that initializes Kura with custom summary models
• Any hardcoded model references

Priority

High - This affects the student experience for the AI Engineering Summit workshop.