Frequently Asked Questions

How does it work?

We have a large crowd of over 10 million users responding to our Rapids (our micro tasks).They are spread all over the world, so we can get human input from over 170 countries. Due to the immense size, we can parallelize tasks and get near instant results.

What about data quality?

We are engineers from various backgrounds ourselves. We know that high quality data makes or breaks AI models. Quality is a priority for us in all steps, hence we monitor user performance meticulously and filter low quality responses. Further, we collect multiple responses for each data point, ensuring high accuracy. Feel free to see for yourself with the free credits provided upon signup.

How do I get started?

There are two ways. The first option is to head over to our self-serve platform, but keep in mind it is still in beta. Second, we love to get to know our users, so please do not hesitate to reach out to us at!

How much does it cost?

We offer a credit based pay-as-you-go model. You can buy credits on our platform and use them for any Rapid type and spend them at your own pace.

For example, let‘s say we have 10 000 generated image pairs from two GenAI models. We would like to conduct a human evaluation for the criteria Visual Aesthetics. To ensure good results, we serve each comparison to 10 different users. In total, this requires 100 000 annotations. Purchasing our Enterprise Package, this would cost you only around 90 USD.

For more info, see our pricing overview.

Can you get the following kind of human feedback?

Not all task types are available on the self-serve platform yet. We have experimented (and keep experimenting) with many more formats, so please let us know your specific requirements at Chances are we already have something for your needs, but only used it internally so far.

More questions?

Please ask us directly at!