Frequently Asked Questions
No, if you choose to use our API on the Zuva hosted cloud then you do not need to run anything locally.
Our API does not have a user interface, it is an API platform that is used as the underlying ML engine for contract analysis. It is designed to be embedded into your solution, therefore it is up to you to determine how to consume and display the data that is returned from the API.
The average time required to process a document depends on the characteristics of the document that you submit. Many factors impact processing time, including page count, number of fields extracted, OCR quality, etc. Our API scales horizontally which allows it to process a document across multiple nodes.
No, the microservices are not dependent on each other, you will have the ability to mix and match the Zuva APIs you need for your development efforts. You can run any processing request after submitting your document(s).
The fields API (with GET method) is used to retrieve a list of all the AI fields available to extract from your document, and their associated field_id. The field_id is then used by the extractions API to extract them from your document.
The 1,300+ out of the box fields were trained on publically available, English documents. However, using the AI Trainer, or the Training API, you are able to train new fields in other latin script languages.
No, the API will not be able to extract fields out of documents that are handwritten.
API usage is managed based on tokens. You are able to create multiple tokens from the admin console. Each token can only access the objects that it created, be it files, or requests.
Our API will delete a document submission after 7 days. You can choose to delete it before 7 days using the delete files API. Once the document has been deleted from Zuva it cannot be restored.
Partially. Every file you submit requires 1 API call. However, the Document Classification/Field Extractions/Language Classification endpoints accept a list of file_ids. In other words, these three endpoints can be used to create multiple requests for multiple files with a single API call.
No, Our API currently does not support zip files.
You can monitor usage through the admin console. Usage is tracked on a per token basis.
Our team of Legal Knowledge Engineers use repositories of publicly-available documents, such as EDGAR, to train AI fields. Our OOTB AI fields are primarily trained on English documents. We also offer a subset of fields in German, and Brazilian Portuguese OOTB. These AI fields are trained using Zuva’s AI Trainer, and then published for use in our API.
Our API is built on Microsoft Azure.
We offer hosting on the Zuva cloud and on premise, the Zuva cloud is hosted in the US and EU.
The OCR Engine used is Kofax Omnipage. Zuva’s OCR pipeline includes additional processing prior to running the results through the machine learning engine.
Training on top of a built-in AI field is possible using the training API, you can use this functionality to alter the performance of the built-in AI fields. This is not a function that is available with AI Trainer, our UI based training product.