Validate your startup idea with proof, not opinions.

From retrieval over 278,288 real records, using TF-IDF with a 160-dimension SVD and a hybrid lexical/semantic blend. No number in a report is generated by a language model. That constraint is the entire point of the product, and it is why the Reports engine loads a real index at startup rather than calling an API.

No — once. It is the only input in the product. The report scores it, the generator builds screens from it, the API engine reads it for the capabilities it implies, and the expert brief is matched on the same text. Change it on the home page and every feature picks up the new one.

No. It is deterministic — rules, templates and an industry classifier trained offline on 75,658 startup descriptions. The same idea produces the same prototype every time, in about 58 milliseconds, with no network call. What you get is a themed set of screens with navigation and flows, exportable as a runnable React + Vite project.

Reports, prototypes and API recommendations are free — generate as many as you like, download them, export the code. The only thing that costs anything is time with a mentor, and the mentor sets that price themselves; many list sessions as free. You see the exact amount before you book, in rupees, with the platform fee shown as a separate line.

No. Every feature runs on its bundled data with no keys set — the API engine falls back to a catalogue of 76 APIs and deterministic extraction, and the other three never call out at all. Adding a language-model key improves the prose in API explanations; it does not change any score.