Challenge
Blog content arrived by email from a distributed network of contributors — engineers, product managers, and external subject-matter experts — each submitting in completely different formats. Some sent structured drafts with clear headings; others sent rambling prose, bullet-pointed notes, or inline-attached Word files with no consistent structure. Every submission required a content editor to manually parse the email, extract the article, reformat it to brand standards, assign metadata, and queue it for publishing. With submission volume increasing and editor capacity fixed, the backlog between article submission and publication was growing to several days.
Approach
We extended the core publishing pipeline to support email as an intake channel. An inbox monitor detects incoming blog submissions based on configurable routing rules — subject-line patterns, sender allowlists, or designated submission addresses. Each email and its attachments are ingested and passed to a Gemini-powered parsing layer that normalizes the content regardless of input format, reconstructing a clean document structure from prose, bullet points, or attached files. The AI extracts the article title, body, metadata signals, and any embedded media, then applies the same categorization and tagging logic used in the Drive-triggered pipeline. The normalized output is assembled into a formatted blog post and published directly to HubSpot CMS via API, with a reply notification sent to the original contributor confirming receipt and publication.
Outcome
Average time from submission email to published post dropped from 2–3 days to under 3 minutes, eliminating the editorial backlog entirely. The system processed over 200 contributor submissions in its first two months without a single manual intervention, handling format variance that would have required individual human judgment on every entry. Contributors reported higher satisfaction with the submission process, and the content team redirected roughly 15 hours per week of formatting work toward higher-value editorial tasks.