Fyn Update: 2020-02
Better document preprocessing, more checks, and UX improvements
We have been adding a few new features to Fyn in these two last weeks, and we're excited to share our progress so far!
Better document preprocessing
Getting a high quality extraction relies heavily on how well we pre-process the documents you send to us. We've been improving how we handle a lot of edge cases in document processing, ranging from handling badly encoded digital documents to improving our processes to increase legibility of badly scanned documents.
We've significantly extended the number of checks we run when extracting the content of a document. These checks give you a good indication of the trustworthiness of the results being generated. Checks range from basic ("can the document that you sent to us be read") to complex ("does the extracted content make logical sense" or "does the company that has sent you this invoice actually exist"). We'll be continuing to add checks in the next weeks and months!
We've done various improvements to the front end, including better check grouping, fixing some strange display behavior in certain browsers, better document pagination, and much more.
We made a large internal change to ensure that all of our code remains "backwards compatible" - in other words, allowing documents that have been parsed using an older version of Fyn to continue to work as expected. This change allows us to continue adding features quickly and keep evolving the platform without causing breaking changes that require significant changes on our customers' ends.