Create strong social media captions right inside your Airtable editorial plan. Marketing teams place a briefing in a new row, and the system writes a ready to use caption that follows your tone and audience notes.

When a new record is created, the Airtable trigger picks it up by the created_at field. The flow waits one minute so the author can finish the Briefing. It then pulls the full record and lets an AI Agent use the briefing plus background notes stored in another Airtable table. The agent runs on an OpenAI chat model, keeps context per record with memory, and returns a clean caption. A Set step maps the output, and the update step writes the caption back to a target field while keeping flags in sync.

Setup needs an Airtable base with created_at, Briefing, and a caption field you choose. Connect an OpenAI API key and map the exact table and column names used in your base. Teams can expect faster copy creation, more consistent tone, and fewer edits. This is ideal for content calendars, campaign planning, and ad copy that must follow brand rules.

• Airtable new record trigger watches the created_at field and polls every minute
• One minute wait gives authors time to finish the Briefing field
• Record lookup by id ensures the correct row is enriched and updated
• AI Agent with a focused system prompt built for social captions
• OpenAI chat model generates text using the Briefing as the main input
• Background Info tool lets the agent read brand and audience notes from another Airtable table
• Window Buffer Memory keeps context per record using the record id as session key
• Set step maps the AI output and the update step writes to the chosen caption field

• Reduce caption drafting time from 30 minutes to 1 minute per post
• Automate up to 90 percent of repetitive copy work from briefing to caption
• Improve tone consistency by using stored audience and style notes
• Connect Airtable and OpenAI without copy and paste steps
• Handle up to five times more posts per week with the same team
• Cut revision cycles by 40 percent through clear, formatted output

1. Import the template into n8n: Create a new workflow in n8n, click the three dots menu, select 'Import from File', then 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. Prepare Airtable: Create or open your editorial base. Add fields named created_at, Briefing, and a caption output field such as SoMe_Text_KI. Ensure created_at is set when a row is created.
4. In the n8n credentials manager, create an Airtable Token credential. Generate a personal access token in your Airtable account with read and write access to the target base, then paste it into n8n.
5. In the n8n credentials manager, create an OpenAI API credential. From your OpenAI account, create an API key and paste it into n8n.
6. Open the Airtable Trigger node. In the 'Credential to connect with' dropdown, click Create new credential if needed, select your Airtable Token, then choose the correct base and table. Confirm the trigger field is created_at.
7. Open the Wait node and keep the delay at one minute, or adjust if your team needs more time to finish the Briefing field.
8. Open the Get Airtable Record Data node and confirm the base and table match the trigger. The id expression should read the current record id.
9. Open the AI Agent and OpenAI Chat Model nodes. Select your OpenAI credential. Keep the provided system prompt or adjust tone rules to match your brand.
10. Open the Background Info tool node and point it to the table that holds audience, tone, and brand notes. Make sure the agent has access to this tool.
11. Open the Set node and confirm the mapping from the AI output to a field named SoMe Text or your chosen variable. In the update node, map that value to your Airtable caption field, for example SoMe_Text_KI.
12. Validation: Add a test record in Airtable with created_at and a clear Briefing. Wait two minutes and check the record updates. If the caption field stays empty, verify field names and table ids, then check the n8n execution logs for errors.