I am using FastAPI, Ollama,n8n. Chain of prompts. starting with prompting 1 big authority content with topic and context >> editing and prompting for selective formatted output based on social media (LinkedIn/WhatsApp/X) , mood, temperature >> Final output ready for use. More details : https://worthifyme.com/

Good question! I managed to develop a No cost , Local Content Generator in 20-30 days using freely available AI tools like, Gemeni, ChatGPT, Codex . https://hashnode.com/forums/thread/i-vibe-coded-megamind-content-studio My take is - I could have taken less time if properly done. Now what is Properly Done - Its more about providing guardrails to AI agent , a direction - rules in place helps - creating better strategy by focusing on steps of creation and testing -> and then start providing prompts step by step to build your app. It helps if you have a bit of coding knowledge.

Cool! I would like to use the API in my App: https://blockster.hashnode.dev/why-i-m-building-a-local-first-ai-content-tool-instead-of-using-saas-ai

To answer in short - I am building on CPU only. Defining a custom ollama model with predefined persona. Flow -> User input (Topic +Context) -> raw output from Ollama -> user validation or edition -> Resent to ollama for platform specific formatting with previous edited output -> final post. May be it is adding time to post creation but you can also grow your own model based on your content style which will refine the output. You may find a screenshot of the output of my first prototype in blog -https://blockster.hashnode.dev/lazy-studio-v1