Built an outbound research and drafting pipeline that turns domain search into reviewed LinkedIn posts and articles, with approval required before publication.
A credible LinkedIn presence needed a steady supply of domain-specific posts and articles, while finding sources, choosing a format, and drafting still depended on ad hoc research.
Built a TypeScript pipeline, code-named fetch, that searches configured engineering domains with Exa and recommends post, article, or skip under a rolling seven-day cap of one article. Articles are drafted through OpenRouter. Posts start from an idea prompt, get a short rough draft, and are polished before they enter the queue. A Vercel cron runs the pipeline Wednesday through Friday, with state in Vercel Blob and a Slack notice when items are ready. The CLI and a small review UI can approve, reject, or revise, and the app does not publish to LinkedIn.
Research, drafting, and review share one queue. Approved items are paste-ready, and a rejection or revision stays in the workflow.
Choose the channel that works best for you.