What do you like best about the product?

The most significant Hyperscience value is structured data extraction. We are seeing high accuracy and a high level of automation - especially with fine-tuned production models.
The second thing I like is the speed of mapping each field and the simplicity of adding new structured documents. Only one blank sample is required to add it to the library of documents.

What do you dislike about the product?

Semi-structured extraction is not as smooth as Structured due to the required involvement of humans in multiple steps. The machine often needs help with the Field classification step and Transcription - this leads to stopping the extraction process in two parts.
The second thing is that training in Semi-structured extraction requires a minimum of 400 samples that are not easy to get. This takes time and makes the implementation slower than they are anticipating.
Finally, but most important - no AI Features are available. They are falling back compared to similar IDP tools on the market.

What problems is the product solving and how is that benefiting you?

Hyperscience provides us with a high level of accuracy extraction for structured documents. After the training, the level of automation is on satisfactory level. It allows us to make an automated flow and reduce the involvement of people for tasks that are not cost-efficient.
Hyperscience allows us to bring additional value to the customers by reducing the TAT and reducing HC (saving unnecessary Ops cost).