Quick overview
Workflow runs at 08:00 (UK), AI stories from TechCrunch, The Verge, and the OpenAI Blog via RSS, AI to select and editorially shape one signal into a post and image prompt, generates and QA-checks a poster image, then publishes to a LinkedIn, logs results to Airtable and Gmail.
How it works
- Runs every day at 08:00 UK time. Reads the latest AI articles from TechCrunch, The Verge, and the OpenAI Blog RSS feeds and merges them into a single list.
- Cleans, deduplicates, and balances the articles by source, then formats the top items into a consolidated text brief.
- Uses OpenAI (GPT-5) to select the strongest operational AI signal and outputs a structured content pack including editorial card fields, or marks the run as “skip” if quality is below threshold. If the signal is publishable, uses OpenAI (GPT-5) to draft, refine, and QA the final LinkedIn post text and derives up to two hashtags.
- Creates an Airtable audit record for the post content and generates an editorial poster image using either an external screenshot service or the OpenAI Images API, then QA-checks the image with OpenAI and retries on failure.
- Uploads the approved image to LinkedIn, publishes the Company Page post, updates Airtable with the LinkedIn URN and URL, and emails a posting log via Gmail.
Setup
- Provide OpenAI credentials (OPENAI_API_KEY) and select them in the OpenAI text and image/QA HTTP request nodes.
- Set Airtable access (AIRTABLE_API_TOKEN) and configure AIRTABLE_BASE_ID and AIRTABLE_TABLE_ID for where the audit records are stored.
- Connect LinkedIn Community Management OAuth2 credentials and set LINKEDIN_ORGANIZATION_URN for the Company Page that publishes posts.
- Connect Gmail OAuth2 credentials and set NOTIFICATION_EMAILS to receive the daily status email.
Additional info
If you want HTML-to-PNG rendering, configure SCREENSHOT_SERVICE_URL (and optionally SCREENSHOT_SERVICE_TOKEN), otherwise rely on the OpenAI image-generation fallback.