Yep, that's exactly why we're building it. Our goal is to make generative output explainable and interpretable. Right now we're doing this by showing the retrieved context, which is fed directly to the LLM to perform generation. You can then verify for yourself whether or not the generated content is hallucinated from the retrived document