Why are current AI models so bad at product idea research? Am I doing it wrong?
retained text
I vibe coded a small agentic code suite, that scrapes the web (webshops, forums, reddit, etc.) looking for user problems and possible hardware product solutions. AI models (Deepseek thinking and ChatGPT Sol) then analyze the data and create hardware product proposals out of the user problems. Why is it that after 10 hours of data collection and various agent refinement rounds, I am still unable to get anything nearly useful? How should one tackle the problem?
document fetched 2026-11-16 00:20 · canonical https://reddit.com/r/ChatGPT/comments/1xwtn2k/why_are_current_ai_models_so_bad_at_product_idea