AI-powered PDF extraction
Lutra uses AI models to handle diverse PDF formats, and is able to support text and images in their contents. This allows it to accurately extract data even from complex layouts. It can identify fields, types, and even nested information, and convert them into structured formats ready for use in databases, spreadsheets, and other applications.
Common Use Cases
Finance: Automate the extraction of invoice details to simplify bookkeeping and payment tracking.
Legal Teams: Extract clauses and deadlines from contracts for better contract management.
Procurement: Organize supplier details and purchase terms from agreements.
Compliance: Extract key regulatory information from documents for audit and reporting purposes.
Data Sources
Lutra can extract data from multiple sources, making it easy to work with PDFs from various locations. Upload a file directly using the 📎 upload function , process PDFs stored in a Google Drive folder, or extract key information from email attachments.
How to use Lutra
Lutra will automatically ask for account permissions when it needs to connect to them, you do not need to configure the integrations beforehand. To see what accounts have been connected to Lutra, you can visit the integrations page.
Upload PDF documents to Lutra using the 📎 upload function, or point it to a Google Drive folder of PDF files. In the chat message, ask Lutra to load one or two example files to start.
"Retrieve the PDF files in this Google drive folder"
"Read this attached PDF file"
Tell Lutra what fields you want to extract, and any requirements about the structure or types.
"Extract all the contract information from this PDF file including renewal dates, payment terms."
Export the results into a Google sheet or other resources
"Save the data you extracted into a new Google sheet"
"Send me an email with a summary of the data you extracted."
Save the process as a playbook to scale and automate it.
Click on the
Create a Playbook button to make a playbook.Ask Lutra to scale up the process on many more emails, or schedule the playbook to automate it.
Tips for Best Results
Start Small: Test Lutra on a small dataset or a few examples to ensure the process runs smoothly before scaling up.
Provide Feedback: Lutra uses AI to learn and improve based on your feedback. Examples of providing feedback:
"You missed the data X, Y, Z. Please try again."
"There are multiple date entries, not just one, please extract all of them."
Review Logs: Check execution logs in the Chat view to understand how Lutra performs tasks. Click log entries to see detailed steps and outcomes.