ONLINEAGENT_OPS 2026.Q3 HOME HUBSMODELSBUILDINGCOMPANIESFILMMAKINGSECURITYLIVING WITH ITSKILLSALL HUBS, ONE TERMINAL → ARTICLES CRAFT RECORD BLOG TOOLS SKILLS SEARCH /
HOME›ARTICLES›HOW TO WRITE AI PROMPTS
ARTICLES · BEGINNER GUIDEBEGINNER

How to Write AI Prompts: A Beginner’s Guide (Prompting 101)

Five steps from a vague request to a useful answer, the four mistakes beginners make most, and starter prompts you can copy for emails, summaries and plans.

READ12 min
WORDS2,331
SECTIONS11
TYPEEXPLAINER
CHECKED25 AUG 26
TL;DR — THE SHORT VERSION

To write a good AI prompt, say what you want, give the background the AI cannot guess, ask for the shape of answer you need, show an example if you have one, and then refine what comes back.

  • The AI only knows what you wrote. Anything you leave out, it fills in for itself.
  • Follow five steps: goal, context, format, example, revise.
  • Avoid the common mistakes: too vague, several jobs at once, no background, and accepting the first answer.
  • Copy a starter prompt from the list below and change the brackets.
  • Image tools work differently: they need subject, setting and camera details, and not all of them read a negative prompt.
◈ STILL THE RIGHT STARTING POINT

Written for beginners. When you are ready for what changed since: context engineering covers why prompting is now the smaller half of the job. For the six elements every good prompt shares, see what makes a good prompt.

To write a good AI prompt, say what you want in plain words, add the background the AI cannot guess, say what shape the answer should take, show an example if you have one, and then refine the answer instead of starting again. A prompt is simply the instruction you type into a chat assistant such as ChatGPT, Claude, Gemini or Copilot.

This guide covers everything you need to start writing prompts that work. No technical background required. The gap between a bad answer and a good one usually comes down to the instruction, and the instruction is the part you control.

What is a prompt?

A prompt is everything the AI model receives before it starts generating. That includes your main instruction, any context you provide, and any constraints you set. Unless memory is switched on in the product you are using, it carries nothing over from earlier sessions. It has no understanding of what you intended if you didn't say it, and no way to ask follow-up questions — unless you build that into the prompt itself.1

This is the core insight that changes how most people approach prompting: the model doesn't know what you mean. It only knows what you wrote. Every detail you leave out is a detail the model will fill in on its own, often in a way that does not suit your purpose. More on what a prompt is.

TAKEAWAY

The model only knows what you wrote. It cannot ask questions, so it fills in anything you leave out, usually wrongly for your purpose.

The most important idea: be specific

Vague prompts produce generic answers. The more specific your instruction, the closer the answer is to what you actually wanted. Anthropic’s documentation offers a simple test: if a colleague with no background would be confused by your prompt, the AI will be too.2

VAGUE PROMPT: “Write an email to my landlord.”

SPECIFIC PROMPT: “Write a short, polite email to my landlord asking for the heating in my flat to be repaired. It stopped working on Monday and I reported it by phone on Tuesday. Ask for a repair date this week. Keep it under 120 words, friendly but firm.”

Same task. The second version says what happened, what you want and how it should sound.

The same rule applies to image tools, where the details are different:

VAGUE IMAGE PROMPT: "a woman in a cafe"

SPECIFIC IMAGE PROMPT: "RAW photo, 1woman, late 20s, ivory blazer, sitting at a marble cafe table, golden hour light from left window, natural skin texture, Canon EOS R5 85mm f/1.2, Vogue editorial style, 8K"

The specific image prompt names the camera, the lighting, the publication reference and the lens. Every named element removes one thing the model would otherwise choose for you.

TAKEAWAY

Vague prompts give generic answers. Say what happened, what you want and how it should sound; for images, name the camera, lens and lighting.

5 steps to your first good prompt

The plain point: each step answers one question the AI would otherwise have to guess. The six elements of a good prompt explains each part in more depth.

1. Say your goal

Start with the job, in one plain sentence: “Write”, “Summarise”, “Plan”, “Explain”, “Compare”. Microsoft’s Copilot guide says a clear goal is the only part a prompt must have.3

2. Add the context

Say who it is for, what the situation is and why you need it. Anthropic says that explaining the reason behind an instruction helps the model give a more targeted answer.2

3. Ask for a format

Say what you want back: a list, a table, an email, three options, a word limit, a tone. Google’s Gemini guide treats the response format as an instruction in its own right.4

4. Show an example

If you have something you like, paste it in and say “in this style”. Google’s Gemini guide recommends always including examples, and calls prompts without them likely to be less effective.4 This is called few-shot prompting.

5. Read the answer and revise

Treat the first answer as a draft. Say what to change (“shorter”, “less formal”, “add a deadline”) rather than starting over. Google’s Workspace guide calls this making it a conversation.5

TAKEAWAY

Goal, context, format, example, revise. Only the goal is required, but each extra step means less guessing, and the second round is usually where a good answer appears.

Common beginner mistakes

The plain point: most disappointing answers trace back to one of four habits.

  • Too vague. “Help me with my CV” could mean a dozen things. Name the job: “Rewrite my CV summary for a junior marketing role, in three sentences.”
  • Several tasks at once. Asking for a plan, a budget, an email and a poster in one message tends to give you four thin answers. Google’s Gemini guide suggests breaking complex requests into simpler parts.4
  • No context. The AI does not know your audience, your deadline or what you have already tried. Google’s guide says to include the information the model needs rather than assume it has it.4
  • Accepting the first answer. Microsoft expects a back-and-forth, and asks users to review and verify what comes back.3 Names, dates, figures and quotes in particular need checking; see hallucination.
TAKEAWAY

One job per prompt, with the background included, and a second round before you accept anything that matters.

Copy-paste starter prompts

The plain point: replace the words in brackets and send. These are this site’s own templates; each one covers goal, context and format.

EMAIL: Write a [polite / firm / friendly] email to [who] about [what happened]. I want them to [what you want]. Keep it under [number] words.
SUMMARY: Summarise the text below for [who], who needs [what they need it for]. Give [3] bullet points, then one sentence on what to do next. [Paste text]
PLAN: Make a [timeframe] plan for [goal]. I have [time, budget, limits]. Show it as a table with one row per [day / week / step].
EXPLAIN: Explain [topic] to [who, e.g. a 14-year-old / a new colleague] in under [number] words, using one everyday example. Then ask me one question to check I understood.
FEEDBACK: Here is my [draft]. It is for [audience] and should [goal]. List the three biggest problems, most important first, and suggest a fix for each. Do not rewrite it.

Prompts for images and video: a 3-part structure

Image and video prompts follow a slightly different pattern. Most follow three parts, whether or not the person writing them knows it:

1. Subject

Who or what is in the picture? Describe the person, object or scene. Put it early. Order matters on at least some models, though vendors differ on what goes first: Google’s Veo formula opens with the camera, for example.6

2. Context and environment

Where is this happening? What is the setting, the mood, the time of day, the lighting and the atmosphere?

3. Constraints and format

What should the output look like? Camera, lens, colour grade, resolution, aspect ratio and style references. Constraints narrow the model’s choices and make results more consistent across runs.

TAKEAWAY

Subject early, then context, then constraints. Order matters on some models, so check your vendor’s formula on what the vendors actually say.

Negative prompts

A negative prompt is a list of things you don't want in an image or video. Check whether your model has one before you write it — this is the detail most beginner guides get wrong. Midjourney has a negative field (--no); Veo has negativePrompt only on Google’s Gemini Enterprise Agent Platform, not in the Gemini API; and Kling has one only on its legacy API — where Kling itself recommends writing negatives inside the prompt. FLUX.2, Runway, Seedance and ChatGPT Images do not, and Runway states that using one “may result in the opposite happening.”7 Where the field exists, it is worth using. Where it does not, the same thinking still applies — you describe your way past the failure instead of naming it. Which model does what.

DOES YOUR MODEL READ A NEGATIVE PROMPT?
Check before you write one. Where there is no field, describe your way past the failure instead.
HAS A NEGATIVE FIELD
Yes: Midjourney
Note: Kling: only on its legacy API, and Kling itself recommends writing negatives inside the prompt
NO NEGATIVE FIELD
No: FLUX.2
No: Runway, which states that using one “may result in the opposite happening.”
No: Seedance
No: ChatGPT Images
Reasoning — summarises this page’s section on negative prompts, page checked 25 Aug 2026; its Kling detail was read in a browser 16 Sep 2026.
NEGATIVE: CGI, plastic skin, airbrushed face, overexposed, watermark, blurry, cartoon, deformed anatomy, extra fingers, flat lighting

On a model that reads one, a negative prompt names the failure modes to avoid. On a model that does not, the same list is still worth writing — as a checklist of what your positive prompt has to rule out. Instead of “no plastic skin” you write “visible pores, uneven natural skin tone.” Same intent, phrased so the model can act on it. In chat assistants the same advice applies: Anthropic suggests saying what to do instead of what not to do.2

TAKEAWAY

Not every model reads a negative prompt. Where the field exists, use it. Where it does not, describe your way past the failure, such as asking for visible pores instead of banning plastic skin.

Why the same prompt gives different results each time

Chat assistants do not give a fixed answer: Microsoft’s Copilot guide warns that the same prompt can produce different responses.3 Image models use randomness as part of the generation process too. A setting called the seed controls this randomness — and by default, it changes every run. That's why two runs of the same prompt produce different outputs. If you find a result you like and want to reproduce it, save the seed number and use it again. Most image models display the seed in the output metadata. See seed meaning.

TAKEAWAY

The seed controls randomness and changes every run by default. Save the seed of a result you like so you can reproduce it.

The iterative workflow

Experienced prompt writers don't write one prompt and expect a perfect output. They write a base prompt, review the output, identify what's wrong, and refine. ChatGPT Images 2.5 makes this particularly straightforward — you can describe changes conversationally ("make her jacket darker, change the background to an interior") and the model adjusts while keeping everything else the same.

For any other model, iteration means updating the prompt text and regenerating. Three to five rounds of refinement is normal for commercial-grade output. First-run perfection is rare and shouldn't be expected.

TAKEAWAY

First-run perfection is rare. Review, spot what is wrong and refine; ChatGPT Images 2.5 lets you do this conversationally, while other models need the prompt text updated.

Questions people ask

Do I need to be polite to AI?

No rule says you must, and the research on whether it changes the answer is mixed. A 2024 study across several languages found rude prompts often did worse but extra politeness did not guarantee better results, while a small 2025 test on one model found the opposite: rude prompts scored higher.89 What every vendor guide does ask for is a clear, specific request, so write in whatever tone you are comfortable with and spend the effort on clarity.

Is a longer prompt better?

Not by itself. Google’s Workspace guide asks for brief but specific language, and Anthropic says precision is what improves the result, so add detail that removes guesswork and cut anything that does not.52

Can I use the same prompt in ChatGPT, Claude and Gemini?

Usually yes for everyday tasks: the vendors’ own guides give the same core advice about clear goals, context, examples and format. OpenAI notes that different kinds of model can need different prompting, so if a prompt matters, try it in each app and adjust.10 Prompting for LLMs covers the differences.

Where to go next

This article covers the foundation. The rest of the site goes deeper into every area:

Related on this site: AI Skills for Work · How to Use ChatGPT · Prompting Techniques Explained · Starter Prompts for Everyday Tasks, by Job · What Is Zero-Shot Prompting? · Free Prompt Checker · Free AI Prompt Library · What Is Prompt Engineering?

The example prompts, starter templates and the five-step order on this page are this site’s own ways of explaining, not claims from these sources.

SOURCES
  1. 1Some products now remember by default: Anthropic, Claude's memory works everywhere, and you decide what's in it, 25 Aug 2026, read at source 16 Sep 2026“Memory is on by default on Free, Pro and Max plans across web, desktop, and mobile.” An earlier version of this paragraph said the model has no memory between sessions. ↑
  2. 2Anthropic, Prompting best practices (Claude developer docs), read at source 5 Oct 2026“Golden rule: Show your prompt to a colleague with minimal context on the task and ask them to follow it. If they’d be confused, Claude will be too.” ↑a“Providing context or motivation behind your instructions, such as explaining to Claude why such behavior is important, can give Claude more information about your goals and help it deliver more targeted responses.” (wording re-read at source 12 Nov 2026) ↑b“Tell Claude what to do instead of what not to do” ↑c“The more precisely you explain what you want, the better the result.” ↑d
  3. 3Microsoft Support, Get started writing prompts in Microsoft Copilot, read at source 5 Oct 2026“All that’s required is a clear goal.” ↑a“Expect some back-and-forth conversation to get the results you’re looking for.” and “Review and verify responses you get from Copilot.” ↑b“Using the same prompt multiple times can result in different responses.” ↑c
  4. 4Google AI for Developers, Prompt design strategies (Gemini API docs), last updated 17 Sep 2026, read at source 5 Oct 2026“You can give instructions that specify the format of the response.” ↑a“We recommend to always include few-shot examples in your prompts.” and “Prompts without few-shot examples are likely to be less effective.” ↑b“For use cases that require complex prompts, you can help the model manage this complexity by breaking things down into simpler components.” ↑c“You can include instructions and information in a prompt that the model needs to solve a problem, instead of assuming that the model has all of the required information.” ↑d
  5. 5Google Workspace, Writing effective AI prompts for business (Prompting guide 101), read at source 5 Oct 2026“Make it a conversation. Fine-tune your prompts if the results don’t meet expectations.” ↑a“Be concise and avoid complexity. State your request in brief — but specific — language.” ↑b
  6. 6Black Forest Labs, FLUX.2 prompting guide, read at source 17 Sep 2026“Word order matters - FLUX.2 pays more attention to what comes first.” An earlier version of this paragraph said all AI models weight earlier tokens more heavily; no source says that for every model. ↑
  7. 7Kling AI API, Text to Video (legacy), read in a browser 16 Sep 2026“It is recommended to supplement negative prompt via negative sentences within positive prompts”. An earlier version of this paragraph listed Kling alongside Midjourney and Veo without qualification. ↑
  8. 8Yin, Wang, Horio, Kawahara & Sekine, Should We Respect LLMs? A Cross-Lingual Study on the Influence of Prompt Politeness on LLM Performance, 2024, read at source 5 Oct 2026“impolite prompts often result in poor performance, but overly polite language does not guarantee better outcomes.” ↑
  9. 9Dobariya & Kumar, Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper), 6 Oct 2025, read at source 5 Oct 2026“Impolite prompts consistently outperformed polite ones” and “These findings differ from earlier studies that associated rudeness with poorer outcomes, suggesting that newer LLMs may respond differently to tonal variation.” ↑
  10. 10OpenAI, Prompt engineering (API docs), read at source 5 Oct 2026“Different model types (like reasoning versus GPT models) might need to be prompted differently to produce the best results.” ↑

Free printable AI cheat sheets, and AI news every week.

Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.

SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK
aipromptgeneer://$ about this siteAI PROMPT GENEERAI explained in plain English. Every claim links to its source, with the date it was checked.NEWSLETTERWhat changed in AI each week, in plain English.SUBSCRIBE FREE ↗AI PROMPT GENEER · MACHINE ASSISTANCE DISCLOSEDEVERY ARTICLE CARRIES ITS OWN CHECKED DATE