Challenge
The client's content team produced blog articles as Word documents but lost hours every week manually reformatting, categorizing, sourcing images, and publishing each post into their CMS. The process was slow, inconsistent, and didn't scale with their publishing cadence.
Approach
We built an automated publishing pipeline. A Word document is dropped into a shared Google Drive folder, which triggers a Python service that passes the content to Google's Gemini model. The AI analyzes and structures the document - extracting title, body, metadata, categories, and tags - and processes associated media (cover and inline images are resized, matched, and uploaded). The structured output is published directly into the HubSpot CMS as a blog post, with the relevant HubDB tables updated in the same run. A notification confirms completion.
Outcome
Documents go from upload to published, fully formatted content in under a minute, with zero manual copy-paste or formatting work. Categorization and tagging are applied consistently by the AI rather than by hand.