I like how accurately Docsumo pulls data from our documents, including ones with different layouts. The review screen is easy to use, with extracted fields next to the original document so we can check and fix values quickly. It was simple to set up new document types and define the fields to extract, and our team got comfortable with the platform quickly. The API connected the output to our existing automation workflows without much effort. Docsumo has cut most of our manual data entry, so our team can focus on reviewing and validating the output and has more time for automation work. The Docsumo team also helped us while we trained the models on our document formats and tuned extraction. Review collected by and hosted on G2.com.
I find that accuracy drops on poor-quality scans and handwritten entries, so those still need more manual review. New or unusual document layouts take a few rounds of training and corrections before extraction becomes reliable. I’d also like more flexible bulk editing in the review screen and more detailed reports on extraction accuracy over time. A confidence score for each field, with low-confidence values sent to a reviewer, and better recognition of handwritten numbers and checkmarks would help my team. Review collected by and hosted on G2.com.



