Am I using ChatGPT the right way for my work?

I've been using ChatGPT as a resource for my work in Union, HR, and Law, aiming to improve worker conditions. My task often involves analyzing PDFs of the Union Contract and Worker Representation Law, making sure to use the latest versions available online. I need the AI to sift through paragraphs and help answer questions like whether a new work model proposed by management is permissible, or what loopholes I should be cautious about. Moreover, I seek arguments for discussions with management. I even created a custom GPT with all these documents. However, I've faced issues in recent months with ChatGPT providing incorrect answers, which has made me doubt my use. I tried out Gemini 2.5, but it seems to have lost its effectiveness too. I recently learned about the potential limitations of different models with PDF interpretations. Should I consider switching to a different model or LLM for this kind of querying?