Spike: Evaluate LLM performance on markdown podcast transcripts

[rjvitorino]

Goal

Test a few LLMs using sample markdown transcripts from the existing website.

Tasks

• Select 2–3 different podcast transcripts already available
• Run them through different models (OpenAI, Ollama, Claude, etc.)
• Capture outputs and evaluate against predefined criteria

[rjvitorino]

Depends on #1 and #2