A team of AI agents that writes your next lead magnet
A multi-agent system that researches, structures, writes by chapters and edits a long-form article — delivering a downloadable lead magnet without writing it by hand.
What it solves
Writing a good lead magnet — a downloadable guide that captures leads on LinkedIn — means researching the topic, structuring a solid outline, drafting each section with depth and editing it all so it reads in a single voice. Done by hand, it’s a days-long process. This system reduces it to typing a topic and waiting.
How it’s built
Several specialized AI agents work in sequence, each with a tightly scoped task:
- Research brief — a first agent turns the topic into five very specific research queries, leaning on a custom knowledge base if the topic requires it.
- Research and outline — a second agent digs into those queries and drafts a complete article outline, citing the sources it consulted.
- Final structure — a third agent turns that outline into the final structure: title, subtitle, introduction, conclusions and a detailed prompt for each chapter.
- Chapter-by-chapter drafting — each chapter is written separately by a team of “writer” agents that know the previous and next chapter, so the text flows without repetition, with sources cited.
- Voice editing — an editor agent stitches all the chapters together, polishes the tone so it reads in a single personal voice, and flags where images should be added.
- Auto-publishing — the result is published to a Google Doc, shared via a public link and ready to distribute.
Why it’s interesting
The relevant pattern isn’t AI writing — anyone can do that — but the role split: each agent has a bounded responsibility and the next one takes its output for granted, much like a human editorial team. Chaining several specialized agents produces better results than asking a single model for everything, and yields a long-form article, well-structured and with its own voice, generated end to end without manual work aside from the final review.
The workflow, in the open
The full workflow is published on GitHub: you can download it, import it into your own n8n and see exactly how it’s built, node by node → lead-magnet-agent.json