BastionGPT supports a GPT-style workflow through its saved prompts and reference data feature, accessible from the right-hand sidebar inside the platform. You can store reusable instructions alongside supporting documents, then call on them whenever you need consistent output for charting, patient education drafts, policy review, or any other recurring task. On the Professional plan ($20 USD per user per month), each saved prompt can include reference documents up to 10,000 words, roughly the equivalent of a 30-page document. The Professional Plus plan ($45 USD per user per month) raises that ceiling to 150,000 words, or about 500 pages, which covers the vast majority of clinical protocols, intake packets, and internal guidelines that teams want their AI assistant to reference.
If you need to attach an enormous reference library to a single prompt, OpenAI's consumer GPTs technically allow larger file uploads. However, they do this by utilizing technology that is more prone to harmful biases and errors (e.g., the "Lost in the Middle" effect or by "chunking" your document and only having the AI review small pieces of your document).
Because BastionGPT is tuned for clinical language and healthcare workflows, the outputs you get from a saved prompt tend to be noticeably more useful than what a general-purpose assistant produces on the same input. Whether you're generating therapy notes, drafting referral letters, summarizing a chart, or translating a discharge summary into plain language, the model handles medical terminology, documentation conventions, and clinical nuance with far more accuracy than a consumer tool. Pair that with enterprise-grade security, and you have an environment where physicians, nurses, therapists, and administrators can actually use AI on real work.
If you'd like a walkthrough of saved prompts in action, or want to talk through a custom build for bigger reference sets, you can schedule time with our team at https://bastiongpt.com/meet.