Turn Airtable into a smart data intake for PDFs. Field descriptions act as prompts, so cells fill themselves with values pulled from each file. Great for teams that read contracts, resumes, or invoices and need clean, structured data fast.

An Airtable webhook sends events when rows or fields change. The flow loads the table schema, finds fields with descriptions, and routes by event type using a switch. For a row change, it updates only the missing fields. For a field change, it updates every row that has a file. The file is downloaded, the PDF text is extracted, and an OpenAI model runs the prompt from the field description to return short answers in the right format. Batches control speed and keep the table responsive while Set and Update nodes write values back to Airtable.

Setup is simple. You need an attachment column for the input file and clear descriptions on the target fields. Add your Airtable token and OpenAI key, publish the webhook, and run the small helper to register Airtable events. Expect a major drop in manual data entry time and fewer errors. Ideal for operations teams that process many documents each week.

• Webhook listener captures row and field change events from Airtable
• Event switch routes updates for single rows or all rows under a changed field
• Table schema is read to find fields with descriptions that act as prompts
• PDF text is extracted from attachment URLs and prepared for the model
• OpenAI generates field values using the field description and type as guidance
• Filters skip rows without files and skip fields that already have values
• Batch processing updates one row at a time for quick visual feedback
• Records are updated in Airtable with a clean mapping built by Set nodes

• Reduce manual data entry from 15 minutes to under 2 minutes per document
• Automate up to 90 percent of repetitive PDF reading for known fields
• Improve data accuracy by using consistent prompts for every row
• Handle 10 times more records with batch updates and routing
• Connect Airtable and OpenAI without custom code

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Airtable and OpenAI. See the Tools Required section above for links to create accounts with these services.
3. In Airtable, create or confirm a base and table with an attachment field for your input files. Add target fields and write clear descriptions that explain what data to extract.
4. In the n8n credentials manager, create a new Airtable Personal Access Token credential. Name it clearly, paste your token from the Airtable API page, and save.
5. In the n8n credentials manager, create a new OpenAI API Key credential. Get the key from your OpenAI account API page, paste it, name it clearly, and save.
6. Open the Airtable nodes in the workflow such as Get Table Schema, Fetch Records, Update Record, and Update Row. In the Credential to connect with dropdown, select your Airtable credential. Set the Base and Table values to match your setup.
7. Open the OpenAI nodes and choose your OpenAI credential. Keep the default prompt style or adjust the system message for your tone and output format.
8. Set the input field name in the helper nodes if needed. Ensure the field that holds your file is named the same in Airtable and in the node settings.
9. Register Airtable webhooks by running the workflow in test mode and executing the mini flow that includes the RecordsChanged Webhook and FieldsChanged Webhook nodes. Confirm that Airtable shows active webhooks.
10. Activate the workflow in n8n so the webhook URL is live. Edit a row in Airtable or create a new field description to trigger an event.
11. Validate results by checking the updated Airtable record. If values do not appear, confirm file URLs are accessible, credential permissions are correct, and the field descriptions are specific.
12. For large tables, use the batch size settings to control throughput and avoid rate limits. If token use is high, simplify prompts or reduce the number of fields updated at once.