Enrich new Airtable influencer prospects with Tubular insights

Build an agent workflow that auto-enriches new influencer prospects in our Airtable base with Tubular Labs intelligence so my team can decide whether to reach out.

Trigger: poll Airtable for a new_record in our "Influencer Prospects" base/table. Use Airtable's first-class new_record poll trigger so the agent fires once per newly added row.

On each fire, the agent should:

1) Read the new Airtable row (Get Record, or List Records scoped to the triggering record id) and extract the creator handle and platform (YouTube / Instagram / TikTok / etc.) plus any other context columns the team filled in.

2) Call Tubular Labs via Execute GraphQL Query against api.tubular.io/graphql to look up that creator. Pull: audience size (followers / subscribers), primary country, age-gender breakdown, top recent videos (title + URL + views), average views per video over the last 30 and 90 days, engagement rate, and the top 3 content categories the creator posts in. If the schema or available fields aren't obvious, the agent should Introspect Schema first to discover what's queryable on this account.

3) Compare the creator against a configurable brand brief (target audience, content themes, must-avoids — surface these as workflow inputs so they're easy to edit) and write a short paragraph "fit assessment" plus a 1-5 fit score.

4) Use Airtable Update Record to write structured fields back onto the same row: audience size, primary geo, engagement rate, top category, fit score (1-5), fit summary (long text), and a status field (e.g. "enriched").

5) If the creator can't be found on Tubular (no result, or GraphQL errors with no usable data), the agent must still call Update Record and set the status field to "not found" with a one-line note explaining why. Never leave a row half-enriched and never fail silently.

Notes for the build: the brand brief, the Airtable base id, the table name/id, the handle column name, and the platform column name should all be workflow inputs the user fills in once at setup so they don't have to edit instructions to retarget a different campaign. Keep the agent stateless per row — one Airtable trigger fire enriches exactly one record.