datasohub – AI-assisted content production in production
From research to a published post in about 7 minutes: a two-model pipeline – Gemini 3 Pro writes, Nano Banana illustrates – with a generative blog system and a data-driven frontend.
At datasohub, content production was entirely manual: research, writing, posting, creating or buying imagery – time-consuming and not scalable. Concept and implementation of the generative blog system and the data-driven frontend were mine.
Tech stack: Gemini 3 Pro (text) · Nano Banana (images) · React · SQLite · cron job
Two models, one pipeline
Two specialised models split the work: Gemini 3 Pro handles research via web search and writes the post, Nano Banana (Google’s Gemini image model) generates the matching imagery. A data-driven frontend with React and SQLite holds it together; the generative blog system assembles text and image into the finished, publishable post.
Seven minutes instead of hours
From research to a published post takes around 7 minutes instead of manual hours – fully automatic via cron job on demand. Live in production at dataso.de/blog.
What I took away from it
The bottleneck shifts from writing to curating: a data-driven frontend keeps the generated content consistent and maintainable. Full automation is technically possible – editorial guardrails ultimately decide the quality.