Enter your Novamine access code, pick an industry, paste a real brief and watch Novamine and the raw model answer it side by side.
Pick your industry, paste a real brief, the kind you would give a consultancy, and watch Novamine and the raw model answer the same challenge side by side, each answer scored for novelty and utility by an independent AI judge. The note beside each control changes with your choice and tells you what to expect before you spend a run.
This preview is by invitation only. Enter the code from your invitation to unlock your runs.
A note on scoring. Official NoveltyBench numbers come only from official runs: the benchmark marks its fixed 1,100-question set with its own trained judges, calibrated on thousands of human judgements. Those instruments can't be pointed at your live brief, so this demo mimics them as faithfully as possible - an independent AI judge from a different model family scores every answer on the same two axes, novelty and utility, for an instant, indicative read.
Every example below comes from a logged run of this exact system: the same brief, answered by the raw model and by Novamini, each answer scored by the independent AI judge (novelty, /10). Condensed here; the full outputs behind any row are available on request.