This is good. But do you think something like this can benchmark forms that are using GenAI like this:

a. Workhack AI forms - https://workhack.ai/?utm_campaign=reachout&utm_medium=dm&utm...

b. Typeform formless forms: https://formless.ai/

Good point. It's logical to expect that the dataset format wouldn't stay the same over time e.g. agents, multimodalities, these forms you brought up. I could see it growing into supporting many common dataset formats or more niche formats if needed.

Platform to help create, augment and iterate LLM eval datasets to be production ready