10 ChatGPT Prompts for Content Creation: Boost Your Workflow
Generic AI content usually isn't an AI problem. It's a structure problem. When creators give ChatGPT a thin prompt, they get thin output back. When they give it a framework, context, constraints, and a clear job, the output gets sharper fast.
That's why the best ChatGPT prompts for content creation don't look like one-line requests. They look like operating systems. One Forbes piece showed that creators can produce a full week of content in a 15-minute ChatGPT session when they use structured prompts that define audience challenges, unique perspectives, expert-backed points, and platform adaptations, rather than asking for random ideas (Forbes on building a content machine with structured prompts). If you're also comparing model behavior, it helps to understand Claude Opus capabilities because prompt structure doesn't translate perfectly across tools.
Use the frameworks below when you want content that sounds deliberate, not generic.
Table of Contents
- 1. The AIDA Framework Prompt Template
- 2. The Role-PlayingPersona Prompt Template
- 3. The 5W1H Five Ws and One H Question Framework Prompt
- 4. The Problem-Solution-Benefit PSB Prompt Template
- 5. The Hook-Story-Offer HSO Framework Prompt
- 6. The Data-Driven Insights Prompt Template
- 7. The SEO-Optimized Content Prompt Template
- 8. The Multi-Format Content Repurposing Prompt Template
- 9. The Specification-Based Constraint Prompt Template
- 10. The Audience-Persona Deep-Dive Prompt Template
- 10 ChatGPT Prompt Templates Comparison
- From Prompts to Production Your Next Steps
1. The AIDA Framework Prompt Template
AIDA is old. It still works because attention, interest, desire, and action are still the path most buyers take before they click, reply, or buy. If your draft feels flat, AIDA usually fixes the order of ideas before it fixes the wording.
This framework works especially well for product pages, launch emails, webinar pages, app store descriptions, and paid social copy. It does less well on educational thought leadership where a hard sell hurts trust.
When AIDA works best
Use AIDA when the reader already has some buying intent but needs help moving forward. A skincare brand can use it for a product page. A SaaS company can use it for a trial signup page. A coach can use it for a workshop landing page.
Practical rule: Don't ask ChatGPT to “write AIDA copy.” Give it the product, the audience, the buying trigger, the objection, and the CTA format.
For teams that want more reliable prompt structure, good generative AI prompt engineering patterns help because AIDA falls apart when you leave too much implied.
Copy and paste prompt
Use this template:
- Role: You are a direct-response copywriter writing for [audience].
- Offer: The product/service is [offer].
- Goal: Write copy that follows AIDA for [asset type].
- Attention: Open with a sharp hook tied to [pain point / desire].
- Interest: Explain why this problem matters now using [context].
- Desire: Show why this offer is different from alternatives such as [competitor or common workaround].
- Action: End with 3 CTA variations for [desired action].
- Constraints: Keep the tone [tone], avoid hype, avoid generic claims, and use concise sentences.
What works:
- audience-specific pain points
- a known objection in the Desire stage
- multiple CTA options
What doesn't:
- broad audiences like “small business owners”
- generic prompts like “make it persuasive”
- forcing AIDA into every content type
2. The Role-PlayingPersona Prompt Template
Most weak AI drafts sound like they were written by “a helpful assistant.” That's the default voice. Role prompts replace that blur with a point of view.
A better prompt doesn't say, “Act like a marketing expert.” It says, “Act like a senior lifecycle marketer at a B2B SaaS company selling to RevOps leaders, with a bias for clarity over brand fluff.” That level of instruction changes examples, vocabulary, and priorities.
Why persona prompts change the draft
A founder writing LinkedIn posts, a support lead drafting help center copy, and a product marketer announcing a feature shouldn't sound the same. Role-playing prompts force the model to make decisions from the right seat.
One overlooked issue is model inconsistency. A 2025 to 2026 industry finding cited by Blaze says 68% of marketers report inconsistent content quality across models due to unoptimized prompts, while only 12% of content creation guides include model-tuned prompt strategies (Blaze on AI-powered social media prompts). If you run multi-model workflows, persona prompts need adjustment per model, not just copy-paste reuse.
Copy and paste prompt
Try this structure in your ChatGPT prompts for content creation:
- Assigned role: Act as a [job title] with [years/level] experience in [industry].
- Audience relationship: You are writing for [specific audience], not the general public.
- Writing priorities: Prioritize [clarity / authority / empathy / brevity].
- Task: Create [asset].
- Voice guardrails: Sound like [traits]. Do not sound like [traits].
- Output format: Deliver [sections / bullets / short paragraphs / script].
Use cases:
- product strategy memo from a product marketing lead
- onboarding email from a customer success manager
- technical explainer from a solutions architect
The trade-off is obvious. Strong personas increase relevance, but they also narrow the draft. That's good when you need precision. It's bad when you're still exploring ideas.
3. The 5W1H Five Ws and One H Question Framework Prompt
When a draft feels incomplete, 5W1H is the fastest repair tool. It forces coverage across who, what, where, when, why, and how. That makes it useful for explainers, reviews, onboarding docs, release notes, and educational posts.
This isn't a flashy framework. It is a practical one. It prevents the common AI failure where the output sounds polished but skips the questions a reader has.
Where this framework beats freestyle prompting
A product team announcing a new analytics dashboard can use 5W1H to avoid vague launch copy. “What does it do?” is not enough. Users also need to know who it's for, when to use it, why it matters, and how to get started.
If you need coverage before elegance, 5W1H is safer than asking for a “comprehensive article.”
It's also strong for search-driven content because it naturally surfaces subtopics people search for in different ways. A freelance educator writing “how to use Notion for research notes” can use it to uncover setup steps, use cases, audience fit, timing, and constraints before polishing the prose.
Copy and paste prompt
Use this prompt:
- Task: Create a draft about [topic] using 5W1H.
- Who: Identify who this is for and who should ignore it.
- What: Define the core concept or product clearly.
- Where: Explain where it applies, including platforms, contexts, or environments.
- When: Explain when someone should use this, buy this, or care about this.
- Why: Explain why it matters in practical terms.
- How: Show the process step by step.
- Constraints: Use plain language, remove repetition, and flag any unanswered questions.
What works well:
- onboarding articles
- comparison content
- internal documentation
- thought pieces that need stronger structure
What doesn't:
- highly emotional storytelling
- short-form ad copy
- founder posts where tension matters more than completeness
4. The Problem-Solution-Benefit PSB Prompt Template
PSB is one of the cleanest frameworks for conversion-focused content. It works because buyers rarely care about your feature list first. They care about the problem they're dealing with, whether your solution feels credible, and what changes after they adopt it.
A lot of ChatGPT output skips the first step. It jumps straight to solution mode. That's a mistake because readers often need to feel understood before they trust the recommendation.
Why PSB converts better than feature dumping
A B2B software company selling reporting automation shouldn't open with dashboard features. It should open with the problem: manual reporting wastes time, introduces inconsistency, and delays decisions. Then it can present the solution and spell out the benefit in terms the buyer can picture.
This framework is useful for feature launch emails, service pages, landing pages, pitch decks, and case study intros. It's less useful for top-of-funnel educational content where teaching should come before selling.
Copy and paste prompt
Use this template:
- Problem: Describe the specific problem faced by [audience]. Include emotional friction and operational friction.
- Solution: Present [product/service/approach] as the response. Explain how it solves the problem without hype.
- Benefit: Translate the solution into real outcomes for [audience].
- Proof cues: Include credible specifics only if provided. If data is missing, keep claims qualitative.
- CTA: End with one clear next step.
A good PSB prompt also includes objections:
- “What might make this audience doubt the solution?”
- “What alternatives are they using now?”
- “What would make the benefit feel believable?”
That last part matters. ChatGPT often writes “save time and grow faster” because nobody forced it to define benefit precisely. Make it earn the claim.
5. The Hook-Story-Offer HSO Framework Prompt
HSO is one of the best frameworks for social media and founder-led content because it matches how people read. They stop for a hook, stay for a story, and act when the offer feels connected rather than bolted on.
The weakness is obvious too. If the hook is fake drama or the story is generic, the whole thing collapses.
What makes HSO work
The best hook isn't always the loudest one. It's the one that creates tension the right audience recognizes. The story then resolves that tension with a concrete experience, lesson, or pattern. The offer comes last and should feel like the natural next step.
A 2026 Forbes analysis cited by Lettercrafted says content with contrarian or emotional hooks generates 2.4x higher engagement and 1.8x more shares than neutral content, yet only 9% of prompt libraries include dedicated prompts for that use case (Lettercrafted on content creation prompts). That's why HSO works best when the hook has a real angle, not recycled motivation language.
Copy and paste prompt
Use this prompt:
- Hook: Generate 10 hooks for [topic]. Make them specific, emotionally resonant, or mildly contrarian without sounding manipulative.
- Story: Build a short story around [event, insight, customer scenario, or founder lesson]. Include a clear tension point and resolution.
- Offer: End with a relevant call to action for [newsletter / demo / comment / download].
- Platform: Adapt for [LinkedIn / X / Instagram / email].
- Constraints: Keep the voice [tone], avoid generic inspiration, and remove filler.
A few real uses:
- founder post about a failed launch that led to a better positioning strategy
- email intro for a webinar
- Instagram carousel built from a customer pain point
- podcast episode description with a stronger opening
6. The Data-Driven Insights Prompt Template
Data-backed content can raise trust fast. It can also subtly damage a piece if the model fills gaps with numbers that sound plausible but cannot be verified.
That trade-off matters because this framework is not just another one-shot prompt. It is a controlled production system for research-backed content. You give ChatGPT the evidence, define the boundaries, and force it to build from approved inputs only. That is how you get strategic content instead of polished guesswork.
I use this template for any asset where credibility matters more than speed. Investor notes, benchmark summaries, analyst-style posts, and executive briefings all fit here. If the source material is weak, this framework exposes the weakness immediately.
Where this framework helps, and where it breaks
The upside is clear. A structured data prompt helps ChatGPT turn raw research into readable content without drifting off-message. It is especially useful when you pair verified inputs with a clear format and channel constraints. Teams using AI content creation tools for structured workflows usually get better output because the model is working from approved context instead of trying to improvise expertise.
The failure mode is quieter. The draft reads well, the phrasing sounds confident, and one unsupported stat can slip into the middle of an otherwise solid piece. That creates extra fact-checking work and can undermine the entire asset.
Required rule: Never let ChatGPT supply numbers for a final draft unless you are checking every claim against the source.
If you want stronger source inputs before prompting, first improve your site's search terms and gather the actual queries, pages, and performance notes tied to the topic. Better source material produces better analysis.
Copy and paste prompt
Use this structure:
- Task: Write [asset type] using only the facts and data below.
- Approved data: [paste verified numbers, quotes, dates, findings, and source notes]
- Rules: Do not invent statistics, quotes, examples, or findings. If support is missing, write a qualitative statement and mark the gap.
- Audience: [audience]
- Angle: [insight, takeaway, or argument the piece should support]
- Tone: [tone]
- Output: [format]
Best use cases:
- investor updates
- benchmark summaries
- trend reports
- executive briefs
- research-backed social posts
A practical tip. Include source notes beside each number, even if they are brief. ChatGPT handles data prompts better when every claim has a visible origin and the prompt makes the limits explicit.
If you do not have clean source material, skip the data angle and use a qualitative framework instead.
7. The SEO-Optimized Content Prompt Template
SEO prompts fail when people ask for “an optimized article” and stop there. That usually produces stiff keyword repetition, vague headings, and filler sections that nobody wants to read.
Search content gets better when you tell the model the query, the intent, the audience, the stage of awareness, and the specific subtopics the page must cover. Good SEO prompting is less about keywords alone and more about matching intent cleanly.
How to stop AI from writing search sludge
A technical SaaS article targeting “API rate limiting best practices” shouldn't sound like a generic marketing blog. It should answer implementation questions, define terms plainly, and include examples a developer can use. A local service page should do something different. Same prompt tool, different architecture.
If you want to sharpen your process, these ChatGPT prompts for marketing are useful as a companion because search content often overlaps with broader messaging strategy. It also helps to improve your site's search terms before prompting so the model is working from real targets, not guesses.
Copy and paste prompt
Use this template:
- Primary keyword: [target keyword]
- Search intent: [informational / commercial / navigational]
- Audience: [specific reader]
- Goal: Create an outline and draft for [asset type]
- Required subtopics: [list]
- On-page requirements: Include a concise intro, scannable H2s, a direct answer near the top, and natural semantic variations.
- Style constraints: Avoid keyword stuffing, avoid generic filler, and write with practical specificity.
- SERP differentiation: Include one angle competitors usually miss.
A practical example:
- primary keyword for a B2B team
- commercial intent
- audience is operations managers
- draft includes comparison criteria, implementation notes, and CTA paths
That's how ChatGPT prompts for content creation become useful for SEO instead of producing search-shaped mush.
8. The Multi-Format Content Repurposing Prompt Template
Repurposing is one of the most effective uses of ChatGPT. A solid source asset can become a week of channel-specific content, but only if the prompt controls message hierarchy, format rules, and CTA by platform.
That distinction matters. Repurposing is not copying a blog post into five shorter versions. It is a structured production workflow that keeps the strategic spine intact while changing the packaging.
A useful toolset for this kind of workflow is a dedicated AI content creation tool library, especially when multiple people need to reuse the same prompt pattern.
Repurposing works when the source asset is strong
Start with something that already has a clear point of view. Good source material includes:
- webinar transcripts with a clear thesis
- blog posts with a defined argument
- customer interview notes
- launch memos
- founder voice notes with strong language you want to preserve
Weak inputs create flat outputs. If the original piece rambles, every derivative asset will need cleanup.
Content Marketing Institute has covered the operational side of this process in its guidance on repurposing content across channels while adapting the asset to each platform's format and audience expectations (Content Marketing Institute on content repurposing). That is the right model for prompting too. Keep the message stable. Change the execution.
Before you write the prompt, lock four variables:
- Core message: the one idea every version must carry
- Audience: who the asset is for
- Claim hierarchy: which points are primary, secondary, and optional
- CTA direction: what action each platform should encourage
This video gives useful context on content repurposing workflow:
Copy and paste prompt
Use this template:
- Source asset: [paste or summarize original content]
- Core message: [single sentence]
- Audience: [audience]
- Format to create: [choose one: LinkedIn post, X thread, Instagram carousel, email intro, short video outline]
- Platform rules: [length, tone, pacing, hook style, CTA style]
- Preserve: [key points, examples, phrases, proof]
- Adapt: [what should change for this platform]
- Avoid: [repetition, generic hooks, platform mismatch, recycled phrasing]
- Output requirement: [final format, number of sections/slides/posts, CTA placement]
What gets better results
Use one format per prompt run. A LinkedIn post and an Instagram carousel usually need different opening logic, different sentence lengths, and different proof selection.
Give ChatGPT the source asset and the job of transformation, not invention. If you want a 7-post thread, say which claim belongs in post one, which example to include in the middle, and what the closing CTA should do.
Set platform-specific trade-offs up front:
- LinkedIn usually rewards a sharper point of view and a cleaner closing takeaway
- X threads need tighter progression and less setup
- Carousels need slide-level clarity, not paragraphs broken into pieces
- Email intros need a reason to keep reading, not a full argument
- Short video outlines need spoken cadence, pattern breaks, and visual cues
Common failure points
The biggest mistake is asking for every derivative asset at once. Quality drops because the model starts averaging formats instead of writing natively for each one.
Other problems show up fast:
- changing the audience midway
- repurposing a draft that never had a sharp message
- keeping the same CTA across every channel
- preserving too much wording from the source, which makes each version feel pasted instead of adapted
Used well, this framework turns ChatGPT into a content production system. You are not asking for one more draft. You are defining how one strategic asset should be rebuilt for each channel without losing the original point.
9. The Specification-Based Constraint Prompt Template
Some teams don't need creativity first. They need compliance, consistency, and low revision volume. That's where constraint prompts win.
This framework is underrated because it looks boring on paper. In practice, it's one of the best ways to make AI usable across support, product documentation, legal-adjacent content, onboarding, and editorial operations.
Constraints are what make outputs reusable
A support team might need answers under a certain length, in a specific tone, with no blame language and no unsupported troubleshooting steps. A documentation team might need every article to include prerequisites, expected outcomes, and a defined command format. If the prompt doesn't specify those rules, the output drifts.
Creatify reports that prompt structures using real performance metrics and audience behavior data instead of generic topic-only requests can improve viral potential and conversion accuracy by 35% to 40% compared with standard prompts, and that this requires a pre-generation workflow that gathers brand voice guidelines, top-performing examples, and platform data (Creatify on social media marketing prompts). Different use case, same principle. Constraints improve outputs because they reduce ambiguity.
Copy and paste prompt
Use this format:
- Task: Write [asset]
- Audience level: [beginner / technical buyer / existing customer]
- Length: [word count or paragraph count]
- Structure: [required sections]
- Tone: [tone]
- Brand rules: [approved language]
- Negative constraints: Do not use [jargon, passive voice, unsupported claims, exclamation-heavy copy]
- Output check: Review the draft against each constraint and revise before finalizing
A strong constraint prompt often outperforms a “creative” prompt because it removes the reasons teams reject AI drafts.
This is especially useful when several writers or departments need similar quality without sounding identical by accident.
10. The Audience-Persona Deep-Dive Prompt Template
Generic prompts produce generic content. If ChatGPT does not know who the reader is, it fills the gaps with average assumptions, broad advice, and weak examples.
That problem shows up fast in conversion-focused work. A founder buying plug-and-play templates wants speed and simplicity. An enterprise IT director reviewing a security vendor wants risk language, implementation detail, and proof. Same topic. Different stakes, objections, and buying criteria.
This framework improves output by defining the reader before a single sentence gets drafted. Out of all the reusable prompt systems in this list, this one does the most to sharpen positioning, relevance, and CTA fit. It turns ChatGPT from a general writer into a segment-aware messaging assistant.
Why persona depth changes the draft
Audience detail changes more than tone. It changes what the model selects, what it leaves out, and how it frames the value.
A weak prompt asks for "a blog post about analytics." A strong persona prompt forces choices such as:
- which problem matters most
- what the reader already knows
- what they distrust
- what proof they need before acting
- what kind of offer makes sense for their budget and role
As noted earlier, repeatable prompts can support productized offers such as templates, guides, and mini-courses. That approach only works when the prompt reflects the buyer's actual context, not a broad market label.
Copy and paste prompt
Use this structure:
- Audience segment: [specific segment]
- Role and context: [job title, company size, experience level, operating environment]
- Primary pain points: [list the top 3 to 5]
- Desired outcomes: [what success looks like]
- Objections: [reasons they might hesitate]
- Current alternatives: [tools, habits, competitors, workarounds]
- Buying triggers: [events or pressures that create urgency]
- Proof required: [examples, case studies, ROI logic, technical detail]
- Tone preference: [direct, analytical, supportive, technical, etc.]
- Task: Create [asset] for this persona
- Rules: Use audience-specific examples, vocabulary, and objections. Remove generic advice meant for broad audiences. End with a CTA that fits this persona's buying stage.
Better inputs produce better positioning
Use real distinctions, not vague labels:
- budget-conscious startup founder trying to validate demand before hiring
- ecommerce operator managing seasonal promotions and margin pressure
- enterprise IT director reviewing vendor risk and procurement hurdles
- course creator packaging educational products for first-time buyers
I use this framework when a draft needs to persuade, not just inform. It takes longer to set up than a one-shot prompt, but the trade-off is worth it. You spend more time defining the reader and less time rewriting generic copy after the fact.
The draft gets sharper because the decisions get sharper. That is the advantage of persona-driven prompting.
10 ChatGPT Prompt Templates Comparison
| Template | Implementation Complexity 🔄 | Resource Requirements ⚡ | Expected Outcomes 📊⭐ | Ideal Use Cases 💡 | Key Advantages ⭐ |
|---|---|---|---|---|---|
| The AIDA Framework Prompt Template | Medium 🔄, structured stages | Low–Medium ⚡, audience insight needed | Higher conversions; measurable 📊 ⭐⭐ | Sales copy, product pages, email campaigns | Proven persuasive structure; easily customizable |
| The Role-Playing/Persona Prompt Template | Low–Medium 🔄, persona setup | Low ⚡, persona definition time | More relevant, authoritative content 📊 ⭐⭐ | Expert voice content, strategy docs, repurposing | Controls tone/voice; speeds credible output |
| The 5W1H Question Framework Prompt | Medium 🔄, systematic prompts | Medium ⚡, research/time for completeness | Comprehensive, fewer gaps 📊 ⭐⭐ | News, tutorials, documentation, explainers | Ensures full coverage; organized flow |
| The Problem-Solution-Benefit (PSB) Prompt Template | Low–Medium 🔄, three-part flow | Low–Medium ⚡, customer insight required | Clear value-driven conversions 📊 ⭐⭐ | Case studies, ads, landing pages, SaaS copy | Outcome-focused; reduces decision friction |
| The Hook-Story-Offer (HSO) Framework Prompt | Medium–High 🔄, narrative crafting | Medium–High ⚡, creative effort and testing | High engagement and shareability 📊 ⭐⭐ | Social media, long-form emails, storytelling | Memorable narratives; strong emotional pull |
| The Data-Driven Insights Prompt Template | High 🔄, citation & analysis | High ⚡, data sourcing & verification | High credibility and authority 📊 ⭐⭐⭐ | Market reports, research, investor content | Research-backed claims; defensible evidence |
| The SEO-Optimized Content Prompt Template | Medium–High 🔄, SEO setup & tuning | Medium ⚡, keyword research & updates | Improved organic visibility & traffic 📊 ⭐⭐ | Blogs, product pages, documentation | SERP-focused; featured-snippet optimization |
| The Multi-Format Content Repurposing Prompt Template | Medium 🔄, format coordination | Medium ⚡, original content + presets | Greater ROI and multi-channel reach 📊 ⭐⭐ | Content scaling, social launches, podcasts | Efficient repurposing; consistent messaging |
| The Specification-Based Constraint Prompt Template | High 🔄, detailed spec writing | Medium–High ⚡, setup and governance | Consistent outputs; fewer revisions 📊 ⭐⭐ | Support docs, API docs, enterprise content | Enforces standards; reduces iteration cycles |
| The Audience-Persona Deep-Dive Prompt Template | High 🔄, deep research & mapping | High ⚡, interviews and validation | Highly targeted, resonant content 📊 ⭐⭐⭐ | Segmented campaigns, product messaging, sales enablement | Precision targeting; improved engagement and conversion |
From Prompts to Production Your Next Steps
The difference between weak AI output and useful AI output is rarely the model alone. It's usually the prompt architecture. Teams that treat ChatGPT like a vending machine get bland drafts. Teams that treat it like a drafting partner, with context, structure, examples, and constraints, get work they can use.
That's also why frameworks beat one-shot prompts. AIDA gives you persuasion structure. PSB gives you a problem-first conversion path. HSO helps with narrative. 5W1H closes information gaps. Constraint prompts reduce revision loops. Persona prompts improve relevance. Repurposing prompts help one idea travel across formats without losing coherence.
There's another practical reason to build framework-based prompts. Prompt quality doesn't transfer perfectly between models. If you're using ChatGPT, Claude, Gemini, Llama, or another system in the same workflow, you need prompt versions that account for how each model handles tone, verbosity, structure, and instruction order. One generic prompt library won't hold up for long if your team is shipping often.
Start small. Pick one real project due this week. Not ten frameworks, just one. If you're writing a landing page, use AIDA or PSB. If you're turning a webinar into social content, use the repurposing template. If your drafts keep missing the reader, use the audience deep-dive template before you ask for any copy.
Then save the winning version. That's the step many users skip. They discover a prompt that works, use it once, and lose it in a notes app or chat history. Teams that get consistent results build a library of reusable prompt frameworks, each tied to a use case, audience, and output format.
A good workflow usually looks like this:
- Choose one framework: Match it to the asset, not to personal preference.
- Load real context: Include audience, offer, examples, voice rules, and constraints.
- Generate a draft: Don't expect perfection from version one.
- Tighten with follow-ups: Ask for stronger hooks, cleaner transitions, shorter sections, or better objections handling.
- Save the final prompt: Reuse it, then refine it again after the next project.
That's where a tool built for prompt operations helps. Prompt Builder is designed for exactly this kind of work. You can generate model-tuned prompts, test variations, refine them in chat, organize what works, and keep reusable templates in one place instead of rebuilding from scratch every time. If you want another perspective on the broader tooling environment, ProdShort's AI content analysis is a useful companion read.
The ultimate win isn't “more AI content.” It's better content production. Better prompts reduce retries, improve consistency, and let teams spend more time on strategy, editing, and distribution. That's the shift worth making.
Prompt Builder helps turn scattered prompt ideas into a working system. If you want one place to generate, refine, test, and save high-performance prompts across ChatGPT, Claude, Gemini, Llama, Mistral, DeepSeek, Perplexity, Grok, and Cohere, Prompt Builder is built for that workflow.