Generate long-form text content with Manus and Gemini Deep Research

A self-taught creator used Manus together with Gemini 2.5 Pro Deep Research for long-form text content generation work.

What They Did

The poster says they paired Manus with Gemini 2.5 Pro Deep Research and a prompting method they call “fractal iteration,” using hierarchical decomposition, style guides, planning structure, and .md/PDF modules. They used that setup to generate and assemble long-form works, including a 300+ page nonfiction book, a 255-page AI ethics project with chapters assembled after generation, and a 100-story collection, while the person directed the methodology and kept extensive notes.

How It Went

They report producing a 300+ page nonfiction book, half of its first chapter, and 100 short stories totaling over 225,000 words, and say later-produced sections still matched style and voice. In comments, they frame the result as strong for cohesive long output, not for hiding AI-written tells.

Worth Knowing

The main catch is that the poster attributes the results to the methodology plus collaboration with Manus, and says they kept extensive notes while assembling chapters after generation.