This article walks through the process of integrating Impira with your Snowflake warehouse so you can extract unstructured data with Impira and stream it back into Snowflake for further analysis.
To see the Impira x Snowflake integration in action, watch this tutorial video of an example workflow involving invoices and payment data.
Connect Snowflake with Impira
Step 1: Bring data into Impira from Snowflake
After signing into Impira, go to Upload, and From Snowflake.
Enter your Snowflake credentials and stage information.
For the “Path prefix” field, enter the path prefix within the Snowflake stage that’s holding the files you’d like to import.
Choose Mount Snowflake stage. Impira will automatically create a new collection and start to bring files in from Snowflake.
Step 2: Stream data back out to Snowflake from Impira
Open your new collection, choose the drop-down menu by the collection name, and choose Configure automations.
Click the + by Destinations and select Snowflake.
Enter your Snowflake account information.
In the field, “Target table,” enter a name for the new table Impira will automatically create in your Snowflake data warehouse.
Note: While you're in the Configure automations page, you can see the activity of this integration by choosing View under Status.
Create fields and extract data in Impira
Step 1: Train your own machine learning model to extract data
With Impira, you can extract data from unstructured and semi-structured files with just a few clicks. Refer to the article, Getting started: Data extraction, and start at Step 3.
Step 2: Fine-tune your models
After Impira populates your table with data predictions, use the Review workflow to ensure all your data predictions are accurate. This retrains your new machine learning models and helps them be more useful for continued use.
After you’ve extracted the fields you need and your dataset is fully confident and accurate, you can now go back to Snowflake to query this table and query it against existing datasets already within your warehouse.