Our customers need us to extract structured data from a strikingly inconsistent set of documents, and we could not do it without Impira. Meanwhile, they're really attentive to our support needs and receptive to feature requests. Again, there's no alternative to Impira if you're trying to embed document processing in your product. We were trying to make it work with Rossum and paid for a year (and it was horribly expensive) and got nothing in return. We never spent the 100 man hours to annotate 500 documents to train their AI. With Impira, you have a trained AI after annotating a single document. I suppose that's the cherry on top. I'm an engineer, however, and I benefit primarily from how easy it is to embed their service in our system - via API and webhooks, etc. Do yourself a favor and try it out.

Bem implementada extração de dados baseada em chave-valor e modelo

É fácil de usar. Ele complementa a ferramenta de ML que usamos para otimização de custos. Extraímos dados não estruturados, e era um processo manual; até certo ponto, conseguimos minimizar isso.