Thank you for the question. The model isn't fine-tuned for a specific domain. As you guessed, I use a Retrieval-Augmented Generation (RAG) setup to provide relevant domain-specific context at the time of query. This allows the base model to stay general-purpose while still producing responses grounded in investment research. Doing so keeps things lightweight, avoids the cost and complexity of fine-tuning, and still delivers targeted responses when paired with reasonably good source material. Hope this helps.