We created our own model from the ground up , no LLM, LLM are good for general work and tasks. However as soon as you are in a niche field you must train your own model.
One of the biggest challenges of building applications with LLMs is data collection and preparation. LLMs are trained on massive amounts of data, and the quality of the data can have a significant impact on the performance of the model. It is important to carefully curate and prepare the data before training an LLM.