Telegram Voice to Published Article Pipeline with an AI Editor Team
Make
Project's Category
Intro
This one turns a Telegram message, typed or spoken, into a polished, SEO ready article after a whole chain of AI agents has reviewed it.
Problems
Producing a decent article meant someone had to research keywords, structure the piece, tighten the SEO, and proofread it before it was fit to publish. All of that took hours per article, especially when the request came in as a rough voice note.
Solutions
We built a Make scenario that takes a Telegram message, transcribes it if it comes in as a voice file, and passes it through a chain of specialized agents: an audience analyzer, a keyword researcher, a title optimizer, a structure creator, a formatter, an SEO enhancer, a conversion optimizer, a quality controller, a human editor step, and a final checker, all sharing memory through an OpenAI model. Once it clears every stage, the finished article gets stored in Airtable and sent straight back to the requester on Telegram.
Final Thoughts
Splitting the work into that many single purpose agents instead of one big prompt is what keeps the output consistent. Each stage catches something the last one missed, and the human editor step means nothing goes out fully unsupervised.
Telegram Voice to Published Article Pipeline with an AI Editor Team
Make
Project's Category
Intro
This one turns a Telegram message, typed or spoken, into a polished, SEO ready article after a whole chain of AI agents has reviewed it.
Problems
Producing a decent article meant someone had to research keywords, structure the piece, tighten the SEO, and proofread it before it was fit to publish. All of that took hours per article, especially when the request came in as a rough voice note.
Solutions
We built a Make scenario that takes a Telegram message, transcribes it if it comes in as a voice file, and passes it through a chain of specialized agents: an audience analyzer, a keyword researcher, a title optimizer, a structure creator, a formatter, an SEO enhancer, a conversion optimizer, a quality controller, a human editor step, and a final checker, all sharing memory through an OpenAI model. Once it clears every stage, the finished article gets stored in Airtable and sent straight back to the requester on Telegram.
Final Thoughts
Splitting the work into that many single purpose agents instead of one big prompt is what keeps the output consistent. Each stage catches something the last one missed, and the human editor step means nothing goes out fully unsupervised.
Building the automation systems that handles business processes.
Clear Communication
Reliable
Professional
Reliable Quality
Building the automation systems that handles business processes.
Clear Communication
Reliable
Professional
Reliable Quality
