SCG — AI content automation

End-to-end AI content automation for every Raise.Band site: from a keyword to a published, SEO-ready page. The whole content team works in it every day.

Content step: block editor with the AI prompt and model for the page (client data blurred)

problem: Content for a network of sites was written, formatted, linked and uploaded to the CMS by hand — slow, expensive and impossible to scale across sites and languages.
built: The whole web app (≈1,550 of 1,610 commits) and, from 2026, the backend too. A step-by-step pipeline — keywords, topic, outline, content, slug, SEO, FAQ — where every step runs on its own model and prompt; a block editor with 16 custom blocks; articles, categories and FAQ for every site; virtual authors; bulk AI refinement; AI-driven internal linking with similarity checks; a provider and model graph (OpenAI, Anthropic, Google, self-hosted models) with a full request and token audit; prompts per language; Strapi publishing with retries and a log. Backend: FastAPI + Celery pipeline, prompt resolver, publishing.
result: all sitesof Raise.Band on one pipeline7 stepskeyword → published page10+LLMs wired into one graph

React 19TypeScriptAnt DesignRedux ToolkitTipTapReact FlowFastAPICeleryPostgreSQLStrapi

Audit of every AI request with tokens in and out

Need something like this? Let’s talk.

Start a project