Guide · AI Writing
The Best AI Writing Prompts
Frameworks, model picks and copy-ready prompts for every writing task — from blog posts and emails to SEO content, product copy and storytelling.
PromptInFlow AI Guides
General AI Models
Image & Video Models
AI Task Guides
- AI WritingYou are here
- AI MarketingRead guide
- AI CodingComing soon
- AI BusinessComing soon
- AI EducationComing soon
AI writing prompts are not magic spells — they're briefs. The same chat model that produces flat, generic prose for a one-line request will produce sharp, on-brand copy when it's given a role, an audience, a goal and a format to fill. The prompts in this guide are built around that shift: instead of asking ChatGPT, Claude or Gemini to "write a blog post about X," they tell the model who it is, who it's writing for, what the piece needs to accomplish, and exactly how the output should be shaped. The result is a first draft you can actually edit, not one you have to rewrite.
This guide is multi-model on purpose. ChatGPT, Claude and Gemini each have a writing personality — and matching the task to the right model is often a bigger lever than tweaking the prompt itself. We use ChatGPT when we want a confident, structured first draft fast. We use Claude when tone, nuance and long-form coherence matter. We use Gemini when the piece has to be grounded in a real document, URL or dataset. The prompts below note which model each one was tuned on, but every template ports across the three with minor adjustments.
You'll find prompts for the eight writing jobs that show up in almost every content workflow — blog writing, email, copywriting, SEO, social, product descriptions, storytelling and technical writing — plus the writing-specific prompt frameworks that hold up across model releases. The goal isn't to give you a list to copy verbatim. It's to teach you the patterns underneath, so when a new model ships or a prompt stops performing, you can rebuild it in two minutes instead of starting from zero.
Treat this page as a working bench, not a museum. Bookmark it, replace every {placeholder} with something specific to your brand and audience, run the prompt in a fresh chat, and iterate with short follow-ups ("tighten the intro," "cut the metaphors," "add a stronger hook") instead of rewriting the original. The writers who get the most out of AI aren't the ones with the longest prompts — they're the ones who edit fastest.
Editorial standards
How we selected these prompts
Why each prompt earns its place in the library.
Every prompt in this guide had to earn its spot. We tested each one across ChatGPT (GPT-5), Claude (Sonnet 4.5) and Gemini (2.5 Pro) on real writing tasks pulled from our own work and from the Prompt InFlow community — landing pages, newsletter drafts, product copy, SEO outlines, LinkedIn posts. A prompt only ships if it produces a usable first draft on at least two of the three models, adapts cleanly to a new brand or audience without rewriting the structure, and avoids the patterns we know hollow out AI prose (vague adjectives, list-of-everything outputs, hedged conclusions).
Writer-tested
Every template was used to ship something real — a sales page, a blog post, a cold email — before being added. If we wouldn't use it on a client deliverable, it doesn't ship here.
Multi-model
Prompts are written to work across ChatGPT, Claude and Gemini. Where one model clearly wins, we say so explicitly instead of pretending all three are interchangeable.
Editable, not final
Good AI writing prompts produce a 70-80% draft a human finishes — not a polished piece. We optimise for prompts that make the editing pass short, not for prompts that fake perfection.
No filler
We cut prompts that rely on hype words, padding or generic "act as an expert" openings. If a line in the prompt doesn't change the output, it gets removed.
Who this guide is for
This guide is for anyone who writes with AI in the loop — and wants the AI to actually pull its weight.
Content marketers
Blog posts, SEO clusters, newsletters and landing pages that don't read like every other AI draft.
Copywriters
Sales pages, ads and email sequences where voice, rhythm and a tight CTA matter.
Founders & operators
Investor updates, About pages, product copy and announcement posts written under time pressure.
SEO specialists
Outlines, briefs, intros and meta copy that hit search intent without keyword stuffing.
Social media managers
LinkedIn posts, X threads and Instagram captions sized and styled for each platform.
Technical writers
Docs, API references and changelogs that stay accurate when the model is grounded properly.
In this guide
What you'll learn
Skim the highlights, then jump to the section you need.
Which AI model to pick for each writing task
A practical breakdown of when ChatGPT, Claude or Gemini is the strongest tool — by task, length and tone requirements.
Prompt frameworks built for writing
The Role–Audience–Goal–Voice–Format–Constraints pattern and how to apply it to blogs, email, copy and social.
8 task-specific prompt libraries
Copy-ready prompts for blog posts, email, copywriting, SEO, social, product descriptions, storytelling and technical writing.
How to edit AI-generated prose
A repeatable cleanup pass that removes hedging, padding and AI tells in under five minutes per draft.
Framework
How great AI writing prompts work
The six-part anatomy behind every prompt that consistently ships usable output.
The best AI writing prompts share the same six-part structure. Once you can see the parts, you can rebuild any prompt in this guide for your own brand, audience and tone in under a minute. We call it the R-A-G-V-F-C framework: Role, Audience, Goal, Voice, Format, Constraints.
Role
Anchor the model in a specific writing identity. "You are a B2B SaaS copywriter who has written 200+ landing pages for product-led startups" outperforms "You are a writer" every time — it narrows vocabulary, raises the quality bar and pre-empts the generic blog-voice default.
Audience
Be specific about who reads this. "Heads of RevOps at 50–500-person SaaS companies" gives the model real targets for vocabulary, pain points and references. "Business owners" gives it nothing — so it averages.
Goal
Define the outcome the writing has to drive: "book a discovery call," "convince readers to start a free trial," "explain a technical concept to a non-technical exec." Goals beat topics — "goal: book a call" produces sharper email than "topic: our product."
Voice
Describe the tone in three to five concrete attributes ("direct, dry, slightly contrarian, no hype") and one or two banned moves ("no rhetorical questions, no 'in today's fast-paced world'"). Voice is the single biggest lever for output that doesn't sound like ChatGPT.
Format
Spell out the shape: word count, section count, headings, paragraph length, CTA placement. "Return as: 90-word intro, four H2 sections of 120–150 words each, single-sentence CTA" is a brief; "write a blog post" is a wish.
Constraints
List what's off-limits — hedged language, em dashes, list-of-everything paragraphs, claims you can't verify. Constraints prevent the model from defaulting to the safest, blandest version of itself.
Three examples — and why they work
Annotated example — high-converting landing page hero
You are a senior B2B SaaS copywriter who has shipped 100+ landing pages for product-led startups. Audience: {ICP — e.g. ops leaders at 50-500 person SaaS companies evaluating new tooling}. Product: {one-line description}. Goal: get the visitor to click "Start free trial" within 8 seconds of landing. Voice: direct, specific, slightly contrarian; no hype, no rhetorical questions, no "in today's fast-paced world." Format: H1 (max 10 words, includes the outcome not the category), one-sentence subhead (max 22 words, names the audience and the unfair advantage), three-bullet proof row (each bullet starts with a verb), primary CTA, secondary CTA. Constraints: no claims I haven't given you, no adjectives without a concrete noun, no em dashes.Annotated example — long-form thought-leadership blog
You are a writer trained on Stripe's blog and Wait But Why. Audience: technical founders who don't have time for fluff. Goal: argue that {thesis} in a way that earns a share on X. Voice: confident, plain-spoken, examples over abstractions, one strong opinion per section. Format: 1,200-1,500 words, four H2s, one image-prompt suggestion per H2, no bulleted lists, no FAQ section, end with a one-paragraph "so what?" Constraints: every claim either has a real example, a number, or is flagged as an opinion. No hedging ("some experts believe…"). No restating the prompt.Annotated example — cold email that gets a reply
You are a B2B copywriter who has booked 500+ meetings via cold email. Audience: {role + company size + situation}. Goal: book a 15-minute call this week. Voice: peer-to-peer, no flattery, no "hope you're well." Format: subject line (max 5 words, lowercase, no clickbait), 3 short paragraphs (under 110 words total), one specific observation about their company, one proof point with a number, one CTA in the form of a question. Output: subject, body, two alternate subject lines. Constraints: no "I noticed," no "quick question," no "reaching out."Browse by use case
Featured prompt categories
Jump into the library that matches what you're working on — each category includes guidance on when to use it and what good output looks like.
- When to use
- Use these prompts whenever you need a strong first draft of long-form prose — blog posts, essays, newsletters, scripts. They're optimised for clarity of structure and voice rather than raw word count.
- What you can achieve
- On-brand drafts in minutes, with a tight intro, a logical structure and a real point of view — instead of the watery "in today's landscape" voice every AI defaults to.
- Best practices
- Always paste in 2–3 paragraphs of your real voice as a reference — it beats any tone description.
- Ask for the outline first, edit it, then ask for the draft. One extra step, half the rewrites.
- Cap word count explicitly. "Around 1,200 words" produces a tighter draft than no limit at all.
- When to use
- Use these when the writing has to drive a measurable action — click, signup, reply, conversion. Marketing prompts trade nuance for clarity of intent and CTA.
- What you can achieve
- Landing-page heroes, ad copy, lifecycle email and SEO briefs that name the audience, the outcome and the CTA — and skip the filler.
- Best practices
- Always state the funnel stage (cold, warm, retention). The model writes a different sentence for each.
- Give the model one banned phrase list. Cutting 5 clichés lifts perceived quality more than adding 5 features.
- Ask for two or three variants in one prompt — variant testing is where AI copy beats human-only workflows.
- When to use
- Use these for the writing that surrounds a business — investor updates, About pages, internal memos, announcement posts, exec ghostwriting. The bar is credibility, not flair.
- What you can achieve
- Drafts that sound like a competent founder or operator wrote them — specific, calm, numbers where it matters, no marketing hype.
- Best practices
- Feed the model real numbers, real customer names (or anonymised stand-ins) and real timelines. Vagueness in = vagueness out.
- Ask for the "so what" sentence first. If the model can't summarise the point in one line, the long version will wander.
- For exec ghostwriting, paste 3–5 of their real LinkedIn posts as a voice reference and explicitly say "match the rhythm, not just the topic."
Hand-picked
Curated prompts
The prompts we reach for most often — copy, paste, adapt.
High-Converting Cold Sales Email
Write a 4-line cold outreach email with a personalized hook, value prop, social proof and soft CTA.
You are a B2B sales copywriter. Write a 4-sentence cold email to {persona} at {company}. Open with a specific observation about {trigger_event}, then connect it to {value_prop}. Include one short social proof line and a low-friction CTA.
SEO Topic Cluster Builder
Generate a pillar page plus 12 supporting article ideas with target keywords and search intent.
Act as an SEO strategist. For the seed topic {topic}, produce one pillar page idea and 12 supporting articles. For each: working title, primary keyword, search intent, and one internal link target.
At a glance
Quick reference tables
Cheat-sheets for picking the right model and the right prompt style.
Best AI model by writing task
A practical, opinionated breakdown — based on testing the same prompt across ChatGPT (GPT-5), Claude (Sonnet 4.5) and Gemini (2.5 Pro). Pick the model that wins the column, then port the prompt across with minor tweaks.
| Writing task | Best model | Why it wins |
|---|---|---|
| Blog posts (long-form, thought leadership) | Claude | Sustains voice and argument across 1,500+ words without flattening. |
| Blog posts (SEO-led, structured) | ChatGPT | Strongest at hitting briefs, headings and word counts on the first try. |
| Cold email | ChatGPT | Tightest copy under 120 words; best at CTA-as-a-question patterns. |
| Lifecycle / nurture email | Claude | Better tonal control across a 5–7 email sequence. |
| Landing page copy | ChatGPT | Fastest at hero + proof + CTA scaffolding with banned-phrase lists respected. |
| SEO outlines & briefs | Gemini | Long-context + URL grounding — paste the SERP, get a brief shaped by what actually ranks. |
| Social posts (LinkedIn, X) | Claude | Most natural rhythm; least likely to add the AI hype voice. |
| Product descriptions | ChatGPT | Tightest feature-to-benefit translation at small word counts. |
| Storytelling / narrative | Claude | Best at scene, character and tonal nuance over multiple paragraphs. |
| Technical writing / docs | Gemini | Stays accurate when grounded in real source docs; flags what it can't verify. |
Strengths and limitations by model
| Model | Strengths | Watch out for |
|---|---|---|
| ChatGPT (GPT-5) | Fast, structured, follows formatting briefs precisely; great for variant testing. | Defaults to confident, slightly generic voice unless you give it a real reference. |
| Claude (Sonnet 4.5) | Strongest tonal control; long-form coherence; respects nuance and constraint. | Sometimes over-hedges; will add caveats unless you explicitly ban them. |
| Gemini (2.5 Pro) | Unmatched on grounded writing — feed it real docs, URLs, datasets, screenshots. | Without an attached artifact it averages the web; voice can feel flatter than Claude. |
Recommended prompt style by task
| Task | Prompt style that wins | Common failure mode |
|---|---|---|
| Blog post | Outline first, then draft section by section. | One-shot "write me a 1,500-word post" → meandering middle. |
| Single tight brief with banned phrases. | Asking for "a friendly email" → fake warmth + flattery. | |
| Landing copy | Hero + proof + CTA, each with hard limits. | Open-ended prompt → marketing word soup. |
| SEO content | Paste the top 3 SERP results, ask for the gap. | Keyword-stuffed prompt → keyword-stuffed draft. |
| Social post | One idea + platform + word cap. | Multi-idea prompts → no-idea posts. |
| Product copy | Features in, benefits + objections out. | Feature list with no audience → spec-sheet prose. |
| Storytelling | Voice reference + constraint on tropes. | No reference → Hallmark-movie default voice. |
| Technical writing | Attach the source doc; require citations. | No source → confidently wrong details. |
Method
How to write better prompts
Four habits that lift the quality of every prompt you write.
Pick the model before you pick the prompt
Reach for Claude when voice or long-form matters, ChatGPT when speed and structure matter, Gemini when you need the writing grounded in real source material. Matching task to model is a bigger quality lever than any prompt tweak.
Paste your voice instead of describing it
Two real paragraphs of your writing teach a model more than any adjective stack. "Match this voice" + a 200-word sample beats "witty, direct, conversational" every time.
Outline, edit, then draft
Ask for an outline first. Edit it. Then ask for the full draft from that edited outline. You'll cut the editing pass in half and avoid the wandering-middle problem in long pieces.
Ban your AI tells
Keep a personal blocklist ("in today's fast-paced world," "navigate the complexities of," "unleash," "delve into," em dashes if they're a tell). Paste it into every writing prompt. Output quality jumps immediately.
Cap length explicitly
AI models pad. Always set a word count or paragraph count. "Around 800 words, four short sections" produces a tighter draft than no constraint, even if the final piece is longer.
Iterate with surgical follow-ups
Don't rewrite the prompt — give one focused note: "tighten section 2," "replace the third example with a number," "cut the conclusion to one sentence." Three small edits beat one big rewrite.
Edit like a human in five minutes
Cut the first paragraph (it's almost always warm-up). Replace adjectives with examples. Delete every hedged sentence. Read it aloud — anywhere your mouth trips, rewrite. That's the difference between AI-flavoured and shippable.
Avoid
Common mistakes
The patterns that quietly tank output quality.
- Asking for "a blog post about X" instead of stating an audience, goal and angle — the model averages and you get filler.
- Using the same model for every writing task — Claude's voice, ChatGPT's structure and Gemini's grounding solve different problems.
- Stacking adjectives instead of giving a real voice reference — "punchy, witty, conversational" is meaningless without an example.
- Skipping word and section limits — unconstrained prompts produce padded prose every time.
- Letting AI hedges through the edit ("it's important to note," "in many cases," "some would argue") — they're the strongest AI tell in the final draft.
- Trusting AI on facts without grounding — if accuracy matters, attach the source doc and require citations, especially on Gemini.
- Publishing the first draft — AI gets you to 70%; the last 30% is what separates your writing from everyone else's.
FAQ
Frequently asked questions
Quick answers about this guide and the prompts in it.
Keep reading
Related guides
The Best ChatGPT Prompts
A curated, copy-ready library of the highest-leverage ChatGPT prompts — organized by use case and tested across GPT-4o and GPT-5.
The Best Claude Prompts
A curated, copy-ready library of the highest-leverage Claude prompts — engineered for Claude's strengths in long-context reasoning, careful writing and structured analysis.
The Best Gemini Prompts
A curated, copy-ready library of the highest-leverage Google Gemini prompts — built around Gemini's strengths in multimodal reasoning, massive context and live Google data.
The Best AI Marketing Prompts
A curated, model-aware library of marketing prompts that ship — built for the channels and tasks marketers actually run every week.
Compatible models
Works well with
Fresh
Latest prompts
Recently added to the Prompt InFlow library.
High-Converting Cold Sales Email
Write a 4-line cold outreach email with a personalized hook, value prop, social proof and soft CTA.
You are a B2B sales copywriter. Write a 4-sentence cold email to {persona} at {company}. Open with a specific observation about {trigger_event}, then connect it to {value_prop}. Include one short social proof line and a low-friction CTA.
SEO Topic Cluster Builder
Generate a pillar page plus 12 supporting article ideas with target keywords and search intent.
Act as an SEO strategist. For the seed topic {topic}, produce one pillar page idea and 12 supporting articles. For each: working title, primary keyword, search intent, and one internal link target.
Senior Engineer Code Review
Get a staff-engineer-grade review of a diff with risk callouts, simpler alternatives and test gaps.
You are a staff engineer. Review this diff and return: (1) correctness risks, (2) simpler alternatives, (3) missing tests, (4) one nit. Be specific. Diff: {diff}
In summary
Conclusion
The ideas to carry forward — and where to go next.
The single biggest shift in AI writing isn't the next model release — it's treating prompts as briefs and treating the model like a junior writer who works fast. Pick the right model for the job (Claude for voice and long-form, ChatGPT for structured speed, Gemini for grounded accuracy), feed it a real role, a real audience, a real voice reference and hard limits, then edit like a human for five minutes. That workflow consistently produces writing that ships — and it scales across every task in this guide.
Don't copy these prompts verbatim. Use them as templates, swap the {placeholders} for your specifics, and iterate with short follow-ups instead of starting over. The writers getting the most out of AI in 2026 aren't the ones with the cleverest prompts; they're the ones with the fastest feedback loop between draft and edit. Build that loop with the frameworks above, and the prompts almost write themselves.
When you're ready to go deeper on a specific model, the ChatGPT, Claude and Gemini guides cover model-native techniques in detail. To find prompts beyond writing, browse the category library — or open one of the prompt generators to build a writing-ready prompt in seconds.
Continue Learning
Prompt Engineering Learning Path
A guided reading order through every Prompt InFlow flagship guide — start with the model you use most, then layer in topical skills.
- 1
Best ChatGPT Prompts
50+ tested ChatGPT prompts and the R-G-C-C-O-E framework that makes them ship usable output.
Model guide - 2
Best Claude Prompts
Long-context reasoning, on-voice writing and XML-tagged briefs for Claude 3.5 and Claude 4.
Model guide - 3
Best Gemini Prompts
Grounded, multimodal and long-context prompts that lean into Gemini's real-time strengths.
Model guide - 4You are here
Best AI Writing Prompts
Voice, structure and editing — the writing playbook across ChatGPT, Claude and Gemini.
Topical guide - 5
Best AI Marketing Prompts
SEO, content, social, email, ads and brand — model-matched prompts for every channel.
Topical guide
Build prompts faster with Prompt InFlow
Browse 500+ tested prompts across every major AI model — or use our generators to build your own in seconds.