thanks for your feedback! Co-founder here...
EyeQuant isn't replacing eye-tracking studies in the narrow sense - it's a completely new use case:
We predict the first 3-5 seconds, not more - but the results are available instantly via a web service (and as a flat rate.)
This helps you to optimize designs iteratively: you upload a screenshot, get the results and optimize the design accordingly - now the great thing is, that you can re-test your changed version instantly, which wasn't possible before.
In our lab, we do a lot of eye-tracking to just evaluate and the results (and refine the models) - here's a comparison with a large eye-tracking study conducted by Tobii: http://bit.ly/frTdh5
Glad to answer anymore questions.
PS: even if you deleted your comment! :)