The anxiety arrived right on schedule. When ChatGPT burst into public consciousness, the literary world responded with the same instinctive dread that greets every new technology capable of producing sentences: this is the end of writing as we know it.

Novelists worried about obsolescence. Poets wondered who would read verse written by a human when a machine could generate it faster. Screenwriters went on strike, partly over the question of whether algorithms would be credited alongside flesh-and-blood collaborators. The fear was understandable, even predictable. It was also, for a growing number of working writers, misplaced.

Because the writers who have spent the past two years actually using AI tools — not theorising about them, not panicking about them, but sitting down with them daily — have arrived at a different conclusion altogether. The technology is not replacing their craft. It is accelerating it. And the key to making it work is not writing ability alone. It is knowing how to ask the right questions.

The Prompt Is the Skill

In the language of artificial intelligence, a prompt is an instruction — the text you feed into a model to generate a response. A bad prompt produces generic, unusable output. A good prompt produces something that feels like it was written by a collaborator who understands your project, your style and your intent.

For writers, this distinction is everything. Asking ChatGPT to "write a chapter about a detective" will return something bland and formulaic. Asking it to analyse the pacing of a specific scene, identify repetitive sentence structures in a passage, or generate five distinct backstory options for an antagonist whose motivations feel underdeveloped — that is a fundamentally different interaction. The output changes because the input changed. The פרומפט — the prompt — is doing the heavy lifting.

This realisation has spawned an entire discipline. Prompt engineering, as it has come to be known, is now taught in university courses, corporate training programmes and professional development workshops worldwide. But for fiction writers, essayists and authors working in Hebrew, English or any other language, the generic prompt guides designed for marketers and software developers are largely useless. A writer's relationship with AI is specific, creative and deeply tied to the messy, intuitive work of storytelling.

A Library Built for Writers

It is this specificity that makes the approach taken by 10 Books — an Israeli digital education platform focused on writers — worth examining. Rather than offering yet another general-purpose AI course, the platform has built a structured prompt library designed exclusively for authors: dozens of ready-to-use prompts organised by the stages of the writing process itself.

The library is divided into the categories that mirror how a book actually gets made. There are prompts for literary editing — strengthening descriptions, sharpening dialogue, deepening scenes, building tension. There are prompts for character development — creating backstories, designing internal conflicts, building chemistry between characters, refining protagonists and antagonists. Plot construction gets its own section, covering everything from basic story structure to twist generation, subplot development and genre-specific pacing.

Then there are the sections most writers overlook until the manuscript is finished and the panic sets in: linguistic editing prompts for hunting repetition, improving clarity and cleaning overloaded prose, and marketing prompts for writing back-cover copy, crafting launch emails and producing social media content that sounds personal rather than algorithmic.

The design philosophy is pragmatic. These are not prompts that require the writer to understand token limits, temperature settings or system-level instructions. They are copy-and-paste tools — select a task, drop the prompt into ChatGPT, and receive output tailored to the specific creative challenge at hand.

The Dialogue Sharpener

To understand why purpose-built prompts outperform improvised ones, consider a common problem: flat dialogue. Every novelist has written a scene where two characters talk and the exchange reads like an information dump in quotation marks. The characters speak in complete sentences. They explain things to each other that both already know. The rhythm is uniform. The subtext is absent.

A writer working without structured prompts might ask ChatGPT something vague: "Make this dialogue better." The result will be marginally improved at best, because the model has no framework for what "better" means in the context of that scene, those characters and that narrative tension.

A purpose-built prompt approaches the same problem differently. It might instruct the model to identify lines where characters state emotions directly rather than revealing them through behaviour, to flag exchanges where the power dynamic between speakers is unclear, or to suggest subtext-driven alternatives where one character says one thing but means another. The output is not a rewrite — it is a diagnostic. The writer remains the author. The AI becomes the reader who notices what the writer has gone blind to after the fourteenth revision.

The Israeli Writing Market

The 10 Books platform operates within a specific cultural context that shapes its approach. Israeli writers working in Hebrew face a unique set of challenges when adopting AI tools. Most prompt guides, tutorials and best-practice resources are written in English, for English-language models, by people who write in English. Hebrew — with its gendered grammar, right-to-left script and literary traditions that draw from Biblical, modern and colloquial registers simultaneously — requires prompts that account for linguistic nuances the English-language AI community rarely considers.

The platform's prompt library addresses this directly, incorporating prompts designed for Hebrew prose, including character development with distinctly Israeli cultural elements and marketing copy calibrated for the Israeli literary market. For Hebrew-speaking writers who have experimented with ChatGPT and found the output stilted or culturally off-key, the problem was likely not the model. It was the prompt.

What AI Cannot Do

It would be dishonest to discuss AI writing tools without acknowledging their limits, and the writers who use them most effectively are typically the most clear-eyed about what the technology cannot provide.

AI cannot write your book. It cannot generate authentic voice — that idiosyncratic combination of rhythm, vocabulary, perspective and personality that makes one writer's prose distinguishable from another's. It cannot feel what your character feels. It cannot understand why a particular sentence breaks your heart or why a scene works even though it violates every structural rule you were taught. These things remain stubbornly, beautifully human.

What AI can do is remove friction from the mechanical aspects of the creative process. It can brainstorm when you are stuck. It can identify patterns in your prose that you are too close to see. It can generate options you would not have considered. It can handle the marketing tasks that most writers despise but cannot avoid. It can, in short, give you back the time and mental energy to do the part of writing that only you can do.

The writers who are thriving with these tools are not the ones who have surrendered their craft to a machine. They are the ones who have learned to direct it — who understand that the quality of what comes out depends entirely on the quality of what goes in.

A prompt is just a question. But for writers who have learned to ask the right ones, it has become the most useful tool in the shed since the red pen.