Promptify

Why we included this project

Promptify's pitch is that you don't have to rebuild prompt scaffolding every time you want a model to extract something from text. Named entity recognition, classification, question answering, and summarization each map to a short Python call, and the results come back as typed Pydantic objects rather than raw strings, so they drop straight into your existing code. Because the backend routes through LiteLLM, the same task object can hit OpenAI, Anthropic, or a local Ollama model just by changing a model string, which keeps prototyping fast and avoids locking you into one vendor. The evaluation helpers (precision, recall, F1, exact match, ROUGE) plus batch and async processing mean the same code can grow from a notebook experiment into a real pipeline. For anyone doing structured extraction from text who would rather spend time on their data than on prompt construction, this is worth a look.