The machine learning capability is good.

The algorithm helps us to get the right data with less training effort.

The technology seems to be a little bit ahead of the standard NLP software. They use more pattern recognition, rather than basing everything on reading static text. The problem with relying on an unchanging text is that it can depend on things such as the formatting, and where things appear within a document. 

This solution can process complex formats, which is helpful for us because we get a lot of PDFs and they come in many different formats. As an example, a carrier might include a list of their rates in a table, whereas a different carrier will write their rates into a paragraph, and yet another will put them into some sort of graphic. If you only capture text then you will only capture part of that data. Pattern recognition ultimately helps us to circumvent the problems in dealing with format variability in documents and allows us to capture the right data.

When it comes to predicting what it is that a document needs manual intervention, CMR Plus does it pretty well. They have a QA cycle that is usually able to highlight such data.

The content recognition ability, in terms of giving us only the data that we need and nothing else, is pretty good. It has a positive effect on our efficiency and by extension, it impacts our infrastructure costs.

The number of features that are provided is vast. I found that the user interface and the APIs for working with Excel were very helpful to me. The OCR functionality is very good, as well.