AI Agent: Dynamic PDF Data Extraction & Baserow Population with OpenAI

Unlock Automated Data Entry from PDFs with this AI Agent

This AI Agent transforms your Baserow tables into smart, self-populating databases. It listens for changes in your Baserow instance (like new PDF uploads or updated field requirements), intelligently extracts relevant information from those PDFs using OpenAI, and then automatically fills in the corresponding fields in your Baserow table. Imagine uploading an invoice PDF and having key details like invoice number, amount, and due date automatically extracted and entered – this agent makes that possible.

It achieves this AI-driven automation by using the 'description' you set for your Baserow fields as dynamic prompts for an OpenAI Large Language Model (LLM). This means you can define what data to extract simply by describing it in Baserow, without needing to modify the workflow itself.

Key Features & Benefits

• AI-Powered PDF Data Extraction: Leverages OpenAI's LLMs (via Langchain) to understand and extract specific data points from PDF documents.
• Dynamic Prompts via Baserow: Uses your Baserow field descriptions as prompts, allowing you to customize data extraction requirements directly within Baserow.
• Automated Baserow Table Population: Automatically updates Baserow rows with the extracted data, saving significant manual effort.
• Reactive Event-Driven Automation: Triggers on Baserow events such as rows.updated (e.g., a new PDF is added to a 'File' field), field.created, or field.updated.
• Efficient Data Handling: Processes only necessary rows/fields, with options for full table updates when field definitions change.
• Flexible Configuration: Adaptable to various PDF structures and data extraction needs through clear prompt engineering in Baserow.
• Streamlined Document Processing: Ideal for automating tasks like invoice processing, contract data extraction, report summarization, and more.

• An n8n instance (Cloud or self-hosted).
• Baserow account (Cloud or self-hosted).
• Baserow API Token: Create a Database Token in Baserow (Account -> Settings -> Database tokens) with read/write access to the target database.
• OpenAI API Key with access to a suitable model (e.g., gpt-3.5-turbo or gpt-4).

1. Download the n8n workflow JSON file.
2. Import the workflow into your n8n instance.
3. Configure Baserow Credentials: In n8n, create a new 'Header Auth' credential. For the 'Name', use Authorization, and for the 'Value', use Token YOUR_BASEROW_API_TOKEN (replace YOUR_BASEROW_API_TOKEN with your actual token). Select this credential in all HTTP Request nodes interacting with Baserow (nodes named 'Table Fields API', 'List Table API', 'Get Row', 'Update Row', 'Update Row1').
4. Configure OpenAI Credentials: In the 'OpenAI Chat Model' and 'OpenAI Chat Model1' nodes, select or create your OpenAI API credential.
5. Prepare Baserow Table: Create a Baserow table. It must include a 'File' field (type: File) where you will upload PDFs. Add other fields where you want the extracted data to go (e.g., 'Invoice Number', 'Total Amount', 'Client Name').
6. Define Dynamic Prompts: For each field you want the AI to populate, edit its 'Description' in Baserow. This description will be used as the prompt for the AI (e.g., for a 'Client Name' field, the description could be "Extract the full client name from the document.").
7. Configure Baserow Webhook: In Baserow, go to your table, click the three dots (...) menu, select 'Webhooks', and 'Create webhook'.
   • Set 'URL' to the Test URL of the 'Baserow Event' node in your n8n workflow (copy from n8n).
   • Method: POST.
   • Enable 'Use field names instead of IDs'.
   • Under 'Events', select 'Let me choose individual events'.
   • Choose: 'rows.updated', 'field.created', and 'field.updated'.
   • For the 'rows.updated' event, click 'Select specific fields' and choose your 'File' field. This ensures it triggers when a file is added/changed.
   • Save the webhook in Baserow.
8. Activate Workflow: In n8n, ensure the 'Baserow Event' webhook node uses the Production URL once testing is complete. Save and activate the n8n workflow.
9. Test: Upload a PDF to the 'File' field in a Baserow row, or create/update a field's description. Check if Baserow is updated with AI-extracted data.