Behind the scenes it’s an engineering-style failure mode and effects analysis using Cerebras/Qwen 3.8. I did a lot of experiments and for this data performance is similar to GPT/Claude/Gemini, but Cerebras inference is much faster.
It currently works on amazon.com only. I’d love some feedback on its usefulness, especially whether the issues match what you'd actually care about before buying.