airtable-lead-loader

Airtable Lead Loader

Create a fresh Airtable table for a list of leads, with all the columns the rest of the pipeline needs already in place, then batch-load every record. Uses parallel agents because Airtable enforces a one-record-per-create-call limit - serial loading of 500 records would take 8-10 minutes; 4 parallel agents cut that to 2-3 minutes.

When to use

• "Load these leads into Airtable: ".
• "Create a new Airtable table for this scrape".
• Phase 2 of either LinkedIn pipeline (called by the orchestrator).
• You have a JSON list of leads from any source and want them in Airtable with the standard pipeline schema.

Connections I need

• Airtable (database) - Required. I list bases, create the table, and load records via Airtable's REST API through Composio.

If Airtable isn't connected I stop and ask you to connect it from the Integrations tab.

Information I need

• The source file with leads - Required. JSON array of objects. At minimum each row needs profileUrl and fullName. Optional: headline, commentText, reactionCount, location, connectionsCount, experience, education, skills. If missing I ask: "Where's the lead list? Pass me a path to a JSON file or paste the array."
• The Airtable base - Required. If you have only one base, I use it. If you have multiple, I list them and ask which one. If missing I ask: "Which Airtable base should I create the new table in?"
• A table name - Optional. Defaults to LinkedIn {sourceType} - {author} - {YYYY-MM-DD} where sourceType is "Commenters" or "Reactors". Override per call if you have a naming convention.

The table schema

I create the table with these fields. Field types match Airtable's REST API conventions.

Lead identification (always populated by the load):

• Full Name (singleLineText)
• Profile URL (url)
• Headline (singleLineText)
• Source Type (singleSelect: "comment", "reaction")
• Source Post URL (url)
• Source Author (singleLineText)
• Scraped At (dateTime)