Document Intelligence
Extract structured data via natural language from documents, images, and spreadsheets.
Feathery Document Intelligence automatically extracts structured data from your documents, either directly uploaded or automatically pulled from your system of record.
Document intelligence is available on the Business plan. Reach out to sales to get access.
Create an Extraction with queries
Go to the AI tab of your account and click Create Extraction.
Give your extraction a name
(Optional) Choose the extraction type. You'll most likely want to build a custom extraction from scratch, but we also support predefined and optimized extractions for certain types of documents, such as security holdings statements.
(Optional) Select the type of file you're extracting. The two types we support today are images / documents and spreadsheets.
Specify your file source(s), which are the file fields that files will be pulled and extracted from.
(Optional) Specify reviewers for your extraction. They will receive email notifications when new extractions are available for review, and they can review / approve extracted data.
Afterwards, for each piece of data you want to extract from the documents, you will add a new query (in plain English) that tells Feathery AI what to extract. If no suitable value is found, the field will be left blank.
For each query, you can define the following components:
Frequency: this is how many values you want to extract for a specific query.
One Value returns a single value
Multiple Values returns multiple related values that may have multiple occurrences as well. For example, a table where you want to pull multiple columns of all relevant rows.
Page Numbers pulls a list of page numbers that contain the answer to your query. This can be used in a classification pipeline with another extraction
For example, if you have a composite document, the first extraction can identify the type and range of each subdocument. Then you can leverage the page numbers in the second extraction conditionally to only process the specific subdocument with the extraction tuned for that subdocument.
Entity: the information you want to pull from the document (e.g. stock holding, account balance, etc.)
Details (recommended): additional details or constraints about the entity to be queried that allow Document Intelligence to fine-tune relevant data (e.g. only pull securities that are domestic).
Save To: which form or hidden field to save the extracted values to.
Upload or connect documents
You can have Document Intelligence run on files uploaded to your form or connect it to native Feathery integrations like Gmail, S3, Google Drive, Egnyte, and more to automatically pull in new documents uploaded to specific storage destinations.
Trigger extraction from files uploaded to your form
Create a form with a file upload field.
In your form, add a button with the AI Document Extraction action and select the Extraction you previously created. Now, any files uploaded to this field and submitted by your users will automatically trigger the Extraction to run. You can also specify if you want your form fillers to wait until the extraction completes so the form populates with extracted data for review / update.
Trigger extraction from files pulled from your file storage
By enabling an integration on the extraction, you can automatically pull new files from your existing file store without needing to manually upload them to Feathery.
You can also trigger an extraction directly from Feathery's public API.
Last updated
