70 prompts that make AI actually useful

I work as a [YOUR JOB]. My daily tasks are: [YOUR 3 MAIN TASKS]. Keep this in mind for the rest of our conversation. Answer me in plain language, do not pad your answers, and if you use a technical term explain it in one sentence right after. If you understand, confirm in one sentence and list 5 areas where you can help me.

Make this your first message in a new chat. Once the model knows you, the answers get sharply more useful.

Your answer did not work for me. Here is why: [WHY IT DID NOT WORK]. Before you answer again, ask me the 3 questions you need answered to get this right. After I reply, rewrite the answer from scratch.

The most common mistake is giving up after a bad answer. Force the model to ask questions and the second attempt almost always lands.

Summarize the text below through the eyes of a [YOUR JOB]. Use this order: 1) The core idea in one sentence. 2) The 5 points that directly affect me. 3) Three concrete steps I can take this week based on this text. 4) Anything missing or questionable in the text. Text: [PASTE THE TEXT HERE]

Plain 'summarize this' is the weakest use. Tie the summary to your own work and reading turns into deciding.

I know nothing about [SUBJECT]. Draw me a 4 stage learning path: 1) The core concepts I need today, each defined in one sentence. 2) Practical exercises for my first week. 3) The 5 mistakes beginners make most often. 4) How I test what I have learned. Ask me a question at each stage and adapt based on my answer.

The closing question matters. It turns a one way lecture into a guide that matches your level.

I need to make this decision: [DECISION]. My options are [OPTION A] and [OPTION B]. My priorities are [YOUR PRIORITIES]. Give me a comparison table without taking sides. After the table, state the biggest risk of each option. At the end give your own recommendation in one paragraph and explain why.

Ask for the table first and the recommendation last. Ask for advice up front and the model defends whatever came to mind first.

I need to email this person: [WHO]. My goal: [WHAT YOU WANT]. The situation between us: [HISTORY AND SENSITIVE POINT]. Keep the tone [formal / warm / direct but not harsh]. Stay under 150 words. Write me 2 versions, one softer and one more direct. Include subject lines.

Asking for two versions beats hoping for one perfect answer. Choosing is faster than writing.

The image I uploaded is a [INVOICE / CONTRACT / SCREENSHOT / CHART]. Do this: 1) Extract the information into a clean list. 2) Flag the points I should pay attention to. 3) Suggest 3 questions I could ask if something is unclear. Do not invent anything, and if something is not in the image say 'not in the image'.

The line 'do not invent, say it is missing' cuts error rates sharply when reading images.

Here is what I need to do today: [LIST YOUR TASKS]. I have [HOW MANY HOURS] to work. My most productive hours are [TIME RANGE]. Fit these into a realistic day. Rules: put the hardest task in my best hours, leave breaks, and clearly tell me what will not fit so I can move it to tomorrow. Give the plan as a table with time slots.

Without the 'tell me what will not fit' rule the AI crams everything in and the plan is useless.

Rewrite the text below as if explaining it to someone with no legal background. Use these headings: 1) What this document says in short. 2) My obligations. 3) The other party's obligations. 4) Clauses that could put me at risk and why. 5) Questions I should ask a lawyer. Note: you are not a lawyer, remind me of that at the end. Text: [PASTE THE TEXT]

AI is not a legal advisor. This prompt informs you, it does not replace your lawyer.

Teach me [SUBJECT], but do not lecture. Use this method: explain the concept in 3 sentences, then ask me a question. Judge my answer, and if I am wrong show me exactly where. If I am right, raise the difficulty. Repeat until I say stop. Never just praise me, tell me plainly when I am wrong.

The 'never just praise me' line is essential. By default the model leans toward agreeing with you.

Below are notes or a transcript from a meeting. Extract: 1) Decisions made. 2) Who does what by when, as a table. 3) Open items with no decision. 4) Details mentioned that will be forgotten unless tracked. Then write a 120 word summary email I can send to attendees. Notes: [PASTE THE NOTES]

Transcribe the recording first and paste it here. Post meeting work drops to about 10 minutes.

I cannot describe exactly what I want. The result I am after is roughly this: [DESCRIBE THE RESULT]. Write me a professional prompt that will produce it. Before writing it, ask me for whatever information is missing. After writing it, briefly explain what changes if I edit each part.

If you memorize one prompt from this list, make it this one. Having the AI write your prompt is the fastest shortcut there is.

Below are 3 pieces my brand has written before. Extract a brand voice guide: 1) Tone in 5 adjectives. 2) Words we use and words we never use. 3) Sentence length and rhythm. 4) How we address the reader. At the end, write the guide as one copyable block I can reuse in other chats. Texts: [PASTE 3 TEXTS]

Save that block. Paste it at the start of every content chat and your writing starts sounding like one brand.

My core idea: [IDEA]. My audience: [AUDIENCE]. Turn this into 10 different pieces of content. For each give: platform (Instagram / LinkedIn / YouTube / blog), format, headline or opening line, and core message. The 10 should approach the idea from different angles, not repeat each other. At the end tell me which one to start with and why.

The bottleneck in content is not ideas, it is multiplying them. This turns one idea into two weeks of calendar.

Write a [LENGTH] second Reels script. Topic: [TOPIC]. Audience: [AUDIENCE]. Structure it as: a stopping line in the first 2 seconds, then 3 fast facts or steps, then one clear call to action. For each scene write what appears on screen and what is said, separately. Use spoken language, not written prose. List the on screen text captions separately too.

The first 2 seconds decide the whole video. If you do not like it, ask for 5 alternatives of just that part.

Topic: [TOPIC]. Target reader: [WHO]. The question in their head when they search: [QUESTION]. Give me an outline with H2 and H3 headings, one sentence under each on what it covers, and an estimated word count. The outline must answer the reader's question clearly within the first 200 words. Suggest 5 FAQs at the end. No cliché intros, get straight to the point.

Outline first, then write. Asking for 2000 words in one go produces vague, generic text.

The text below reads like it was written by AI. Rewrite it by: removing em dashes and semicolons, cutting filler phrases like 'in today's world' and 'it is important to note', shortening one sentence in every paragraph, adding at least 2 concrete numbers or examples, and breaking up repeated sentence lengths. Do not change the meaning. Text: [PASTE THE TEXT]

What gives AI text away is not vocabulary, it is uniform rhythm. Varying sentence length is the most effective fix.

Write 15 headlines for this content: [SUMMARIZE THE CONTENT]. Five should spark curiosity, five should promise a benefit, five should use a number or list. Keep every one under 60 characters. No hype, no clickbait, do not promise anything the content does not deliver. At the end pick the strongest 3 and explain why.

You ask for 15 not to choose from 15. The first 5 are always clichés, the good ones start around number 8.

Write a LinkedIn post. What I want to share: [THE EVENT OR LESSON]. My audience: [AUDIENCE]. Structure: a first line that stands alone and earns the click, then 3 short paragraphs telling the story, then one question to the reader. No emojis, maximum 3 hashtags. Avoid a bragging tone, put the lesson forward. Stay under 200 words.

On LinkedIn the first line appears alone in the preview. Writing it separately directly affects your click rate.

Build me a [HOW MANY] week content calendar. Platform: [PLATFORM]. [HOW MANY] posts per week. My topics: [YOUR CORE TOPICS]. Rule: each week needs 1 educational, 1 persuasive and 1 personal piece. Table columns: date, format, headline, core message, call to action. At the end, group the pieces that could be produced in a single batch session.

That last line halves production time. Batching similar pieces into one session is the biggest time win there is.

Plan a [LENGTH] minute YouTube video. Topic: [TOPIC]. Viewer: [WHO]. Give me: 1) The exact lines for the first 15 seconds. 2) A section by section flow with timing. 3) The moments a viewer is likely to drop off and what I should do there. 4) Three title options plus a description. 5) At most 4 words for the thumbnail.

Point 3 is what most people skip. Watch time matters more than the title on YouTube.

Adapt the text below for 4 platforms: an Instagram post, a LinkedIn post, an X post, and a newsletter intro. Rewrite for each platform's language and length, do not copy paste. Under each version explain in one sentence why you changed what you changed. Text: [PASTE THE TEXT]

Posting identical text everywhere is the most common mistake. When adapting takes 30 seconds there is no excuse left.

Turn this experience into a story: [WHAT HAPPENED]. Structure: 1) Set the scene so the reader sees where they are. 2) Introduce the problem or tension. 3) Tell what I did. 4) State what I learned in one sentence. Do not exaggerate or add details that did not happen, work only with what I gave you. In the lesson part do not preach, just report what you saw.

'Do not add details' is essential. Otherwise the model decorates the story and puts falsehoods in your mouth.

Below are comments on my content. Write a reply for each. Rules: keep replies to positive comments short and genuine, answer questions directly and point to content when useful, respond to negative comments without getting defensive. No template language, every reply should be different. Maximum 2 sentences each. Comments: [PASTE THE COMMENTS]

Replying to comments lifts reach directly but eats time. This turns 20 comments into 2 minutes.

Below is a piece of my competitor's content that performed well. Analyze it: 1) What emotion or need it targets. 2) How it is structured. 3) Three things that work. 4) What it leaves out or covers superficially. 5) How I could cover the same topic better. Do not suggest copying, suggest differentiating. Content: [PASTE IT]

Point 5 is the one that matters. Filling the gap they left beats imitating what they did.

Write this week's newsletter. Main topic: [TOPIC]. My subscribers: [WHO]. Structure: 1) A personal opening, 2 sentences. 2) The single big idea this week. 3) Three steps the reader can apply immediately. 4) Something that caught my attention this week. 5) One call to action. Stay under 400 words. Do not use sales language, be useful. Give 3 subject line options.

The most skipped rule in newsletters is one call to action. Ask for three things and readers do none of them.

My core topic: [TOPIC]. My audience: [AUDIENCE]. List 15 questions people actually search for on this topic that are not answered well. For each add: search intent (informational / comparison / ready to buy), estimated difficulty, and the unique angle I could bring. No generic questions, be specific.

AI does not know search volume, so verify the list in Google. It is still the fastest way to generate ideas.

Below is a transcript of a conversation. Produce: 1) The 5 strongest quotes, verbatim. 2) Three social posts. 3) One blog outline. 4) The best moments for short video with timestamps. Do not invent quotes, use only sentences that appear in the transcript. Transcript: [PASTE IT]

One recording can feed weeks of content. The moment you realize your problem is distribution and not production, this prompt becomes essential.

Critique the text below like a ruthless editor. Do not praise me at all. Tell me: 1) The exact sentence where the reader gets bored. 2) Where I promise something and fail to deliver. 3) Claims with no evidence. 4) Every sentence that could be cut. 5) Summarize the text in one sentence, and if you cannot, tell me, because that means it has no focus. Text: [PASTE IT]

The model is polite by default. Feedback you get without saying 'do not praise me' is useless.

I need to give a [HOW MANY] minute presentation. Topic: [TOPIC]. Audience: [WHO]. My goal: [WHAT SHOULD HAPPEN]. Give me a slide by slide structure: one idea per slide, at most 15 words on the slide, and my speaker notes separately. Decide the one sentence I want the audience to remember and tie the whole structure to it.

It works even better if you make it decide that one sentence first. The talk gets built around it.

Write me a professional image generation prompt. The image I want: [DESCRIBE IT]. The tool I will use: [Midjourney / DALL-E / Firefly]. Include: subject, composition, lighting, color palette, mood, camera or lens feel, style reference. Give 3 variations at the end: one minimal, one cinematic, one graphic. Briefly explain what each keyword does.

In image generation word choice is everything. Having the model write the prompt is the fastest way to learn.

Suggest a color palette for my brand. Brand: [BRAND AND WHAT IT DOES]. Audience: [AUDIENCE]. The feeling I want: [FEELING]. Give me 3 different palettes. Each with: primary, secondary, accent, dark and light neutrals, all with HEX codes. Explain each choice in one sentence. Also state which text color works on which background for readability.

That last line matters. Most good looking palettes fail in practice because the text cannot be read.

Prepare a logo design brief. Brand: [BRAND]. Industry: [INDUSTRY]. Audience: [AUDIENCE]. Competitors: [COMPETITORS]. Include: the brand's essence in one sentence, clichés to avoid, 3 different directions (each with a name, rationale, and a described visual reference), where the logo will be used and the technical constraints that follow. List the industry clichés explicitly so I can avoid them.

The cliché list is the most valuable part. Knowing your industry's 5 clichés solves half the design problem.

Evaluate the design I uploaded like a professional art director. Cover: 1) Where the eye goes first and whether that is where it should go. 2) Hierarchy problems. 3) Use of whitespace. 4) Typography choices and readability. 5) Color contrast. 6) What breaks on mobile. Give a concrete fix under each heading, no vague 'could be better' notes. No need to praise me.

Point 6 gets skipped in most designs. Finding out late that a desktop perfect layout collapses on mobile is expensive.

Suggest font pairings for my brand. Brand character: [CHARACTER]. Usage: [WEB / PRINT / BOTH]. Budget: [FREE / PAID IS FINE]. Give 4 pairings. For each: heading font, body font, why they work together, whether it supports extended Latin characters, and where to get it. Do not suggest fonts with incomplete character sets, this is a hard rule for me.

The character set warning is essential. The model will not think of it and will suggest fonts missing the letters you need.

My product: [PRODUCT]. Brand feel: [FEEL]. Describe 5 different photo scenes for it. For each: surface and background, light direction and hardness, props, camera angle, color harmony, and which product feature it highlights. Two of the scenes must be shootable on a phone, mark those separately.

The phone requirement brings it down to earth. Studio only ideas usually never get executed.

Design 5 reusable Instagram post templates. Brand colors: [COLORS]. Font: [FONT]. Template types: info card, quote, before and after, product feature, announcement. For each describe: layout (what sits where), type hierarchy and size relationships, spacing rule, color use, and how page two continues in a carousel. Describe it clearly enough that I can build it in Canva.

The 'clear enough to build in Canva' line turns abstract design advice into an executable instruction.

Analyze the image I uploaded and write a detailed prompt so I can produce something similar. Break down: composition, light source and direction, color temperature, texture and material feel, depth, style period or movement. Then turn it into a single generation prompt. Copyright note: I want an original image in the same spirit, not a copy, so do not put any brand or artist name in the prompt.

Leaving out artist names is practical as well as ethical. Many tools now reject those names outright.

Product: [PRODUCT]. Price segment: [BUDGET / MID / LUXURY]. Shelf context: [WHERE IT SELLS]. Suggest 4 packaging directions. For each: material and surface feel, form, color approach, typographic character, what is readable from what distance on shelf, and production difficulty. Also describe what competitors usually do so I can differentiate.

'What is readable from what distance' is the most practical packaging test there is. Brand at 3 meters, product at 1, details in hand.

Structure a [PAGE TYPE] page. Goal: [WHAT THE VISITOR SHOULD DO]. Audience: [WHO]. Break it into sections: the purpose of each, the types of copy inside, image needs, and whether it has a call to action. Rules: what I do must be clear in the first screen, every section must give a reason to reach the next, and describe separately how the order changes on mobile. Ask me questions where you cannot decide.

The 'every section gives a reason for the next' rule turns a list of blocks into an actual flow.

Project: [PROJECT]. Target feeling: [FEELING]. Suggest 3 visual directions. For each: a name, 5 keywords, color approach, typographic character, photography style, what kind of brands use this direction, and its risk. Then compare all three in one table and recommend the best fit for my goal with reasoning.

Asking for each direction's risk is unusual and the most useful part. A direction whose risk you do not know cannot be defended in a pitch.

I will use the images below on my website. For each give: 1) Alt text for SEO and screen readers, under 125 characters, no keyword stuffing. 2) A suggested file name. 3) Any caption that should appear next to it. A blind user listening to the page should lose no information, write with that in mind.

Alt text is both accessibility and SEO work. AI search reads image context from that text too.

My business: [WHAT YOU DO]. My customers so far: [DESCRIBE THEM]. Give me 3 ideal customer profiles. For each: who they are, the problem they have, how they solve it today, why they would switch to me, their biggest objection before buying, and where I can find them. Explicitly mark where you are making assumptions so I can verify them.

Without that last line the model presents guesses as facts. Verify the marked assumptions by asking real customers.

I am writing a proposal for a client. Client: [CLIENT]. Their need: [NEED]. My solution: [SOLUTION]. Timeline and price: [TIMELINE AND PRICE]. Structure it as: 1) Their problem in their own words. 2) What the solution is. 3) Step by step process and deliverables. 4) What is not included. 5) The return on this investment. Put the price last. Do not praise yourself, talk about their outcome.

The 'not included' section is the most valuable part of a proposal. It stops scope creep before it starts.

My service: [SERVICE]. My price: [PRICE]. The objections I hear most: [OBJECTIONS]. For each objection write 3 different responses: one that restates value, one that compares, one that moves forward by asking a question. Also mark which objections actually mean 'I do not see the value' rather than 'it is expensive'. Do not suggest discounts, discounting is a last resort.

Most price objections are value problems. Discounting before you see the difference costs you both margin and position.

I am reaching out to this prospect for the first time: [COMPANY AND WHAT THEY DO]. The concrete opportunity I see: [OPPORTUNITY]. What I do: [YOUR WORK]. Write a short message. Rules: under 80 words, the first sentence must be about them not me, share one concrete observation, and end with a small ask (not a meeting). No template sounding sentences. Give 3 versions.

The small ask matters. Asking for a meeting in the first message lowers reply rates, asking one question raises them.

My industry: [INDUSTRY]. My competitors: [COMPETITORS]. For each, extract what you can: positioning statement, target customer, price perception, where they are strong, where they are weak. Then build a table and mark the gaps. At the end give me 3 positioning options. Mark anything you are unsure about as 'needs verification'.

The model can be wrong about current company details. Verify the table on their websites, but let it build the analysis skeleton.

My service: [SERVICE]. My cost: [COST]. My current price: [PRICE]. My customer: [WHO]. Build me a 3 tier pricing structure. For each tier: who it is for, what is included, what is not, and the pricing logic. Make the middle tier the most attractive and explain how you did it. At the end, tell me which customer segment I lose and which I gain if I raise prices.

Three tiers exist to create a basis for comparison, not to offer choice. A single price looks expensive, the middle of three looks reasonable.

A customer sent this complaint: [COMPLAINT]. What actually happened on my side: [YOUR SIDE]. Write a response. Rules: do not get defensive, acknowledge their experience first, then explain what happened, then offer a concrete resolution, then say what you will do so it does not recur. No excuses, no repeated apologies. Give two versions, one formal and one warm.

Order matters. Explain first and they hear defense, acknowledge first and they hear a solution.

My business idea: [IDEA]. Target customer: [WHO]. Interrogate this idea ruthlessly. Give me: 1) The 5 most likely reasons it fails. 2) Things I am assuming but have not proven. 3) The cheapest way to test those assumptions in 2 weeks. 4) Who is already in this market. 5) What finding would convince me to kill it today. Do not encourage me, be realistic.

Question 5 is the valuable one. Defining the kill criterion up front saves months of walking the wrong road.

I do this task every time and I need to delegate it: [TASK]. I will describe how I do it step by step, and you turn it into a procedure someone else can follow. Include: prerequisites, steps, decision points at each step, common mistakes, and how we know the job is done. Ask me about anything I left out. Here is how I do it: [DESCRIBE]

What blocks delegation is usually not time, it is getting what is in your head onto paper. This solves exactly that.

Tomorrow I meet this prospect: [PROSPECT AND THEIR BUSINESS]. Goal of the meeting: [GOAL]. Prepare me: 1) The 8 questions I should ask, in priority order. 2) Their 5 likely objections with short answers. 3) Things I should never say in this meeting. 4) How I can tell it is going well. 5) The single closing sentence. Focus me on asking questions, not on presenting.

That last instruction changes the balance of the call. The side that listens wins the sales conversation.

Product or service: [WHAT]. Audience: [WHO]. Platform: [Google / Meta / LinkedIn]. Budget level: [BUDGET]. Write 6 ad variations. Two emphasizing the problem, two the outcome, two answering an objection. Separate headline and description for each, respecting the platform's character limits. No exaggerated promises. At the end suggest which copy to show which audience.

Run all six at once and decide after you have data. Nobody knows in advance which one lands.

My business: [BUSINESS]. Where customers first hear about me: [CHANNEL]. Buying process: [PROCESS]. Map the customer journey: awareness, consideration, decision, purchase, after. At each stage: what the customer thinks, what worries them, what they expect from me, what I need to do, and what is missing today. Mark the biggest drop off point and tell me to focus there first.

Trying to improve every stage at once is the classic mistake. Fixing the single drop off point usually pays the most.

This week's data: [PASTE THE DATA]. Prepare a weekly report: 1) Status at a glance, 3 bullets. 2) What changed versus last week and likely causes. 3) Early warning signals to watch. 4) Three actions for this week in priority order. When interpreting numbers, mark anything you are not sure about as 'hypothesis'. Suggest which charts to build.

The 'hypothesis' marker is essential. AI invents causes from data quickly, and verifying which are real is your job.

I am hiring for this role: [ROLE]. What the job actually involves day to day: [WORK]. The person I am looking for: [TRAITS]. Give me: 1) A job post with no clichés that describes the work honestly. 2) Eight interview questions, half technical and half behavioral. 3) What a good answer to each sounds like. 4) A description of what failure looks like in this role.

Point 4 sharpens the hiring decision. You cannot measure success without defining failure.

Below is feedback from my customers. Analyze it: 1) Recurring themes and how often each appears. 2) What gets praised. 3) What gets complained about. 4) What nobody says directly but you read between the lines. 5) The single change I should make based on this. Keep your interpretation separate from their words. Feedback: [PASTE IT]

Point 4 is the valuable one. Customers usually imply problems rather than state them.

My business: [BUSINESS]. Team size: [HOW MANY]. My recurring weekly tasks: [LIST THEM]. Put into a table which of these AI can speed up and which must stay human. Each row: task, current time, estimated time with AI, which tool, the risk, and the first step I should take. Do not exaggerate, mark the ones where it makes no real difference as 'keep as is'.

The 'keep as is' column is the honest part of this table. Handing everything to AI costs quality, not gains efficiency.

I want to build this: [DESCRIBE YOUR IDEA]. I cannot code at all. Give me: 1) The simplest way to do it and which tool. 2) Why that tool and why not the alternatives. 3) A step by step path from zero to working. 4) Where I am likely to get stuck and what to do there. 5) How long it takes and what I need to pay for. Explain any technical term right where you use it.

The biggest beginner mistake is choosing the tool alone. Ask this first, then start building.

I am getting this error: [PASTE THE ERROR]. What I was trying to do: [WHAT YOU WERE DOING]. Tool I am using: [TOOL]. Give me: 1) What this error means in plain language. 2) The 3 most likely causes in order of probability. 3) How to check each one. 4) The fix steps. Say so if you are guessing. Do not make me rewrite everything, show me the smallest fix.

Say 'show me the smallest fix' or the model rewrites your working code and introduces new errors.

I need to do this in Excel or Google Sheets: [WHAT YOU WANT]. My data is laid out like this: [DESCRIBE THE COLUMNS]. Write the formula, then explain what each part does. Also tell me whether the formula breaks if the data changes. Give the function names for my locale and put the English equivalent in parentheses.

The locale note saves time. Otherwise you fight a broken formula without knowing why.

Every [DAY / WEEK] I do this manually: [DESCRIBE IT STEP BY STEP]. I want to automate it. Give me: 1) The 3 best approaches, simplest to most complex. 2) Setup time and cost for each. 3) Pick the one that fits my level and walk me through it. 4) How I will know if the automation breaks. My technical level: [LEVEL]. Describe the setup click by click.

Point 4 is missing from most automation guides. An automation that silently stops working is worse than none.

Explain the code below to me. My programming level: [LEVEL]. In this order: 1) What this code does in one sentence. 2) What happens block by block, not line by line. 3) The parts I can safely change. 4) The parts that are dangerous to change. 5) Anything you find missing or risky. Code: [PASTE THE CODE]

Points 3 and 4 are the most useful information when coding with AI. Knowing what you can touch is freeing.

I am having a website built, or building it with AI. My business: [BUSINESS]. Purpose of the site: [WHAT THE VISITOR SHOULD DO]. Write me a non technical but complete requirements list: pages, what belongs on each, forms and integrations, mobile expectations, speed and SEO requirements, and what I should check at handover. Write it clearly enough to hand straight to a designer or an AI.

This list works the same whether you hire a freelancer or build with AI. A vague brief is the most expensive mistake.

I have messy data: [PASTE OR DESCRIBE IT]. Do this: 1) Find duplicates. 2) List formatting inconsistencies (dates, phone numbers, casing). 3) Identify missing fields. 4) Give me the cleaned version as a table. 5) Tell me what to set up so this cleaning happens automatically next time. Also list everything you deleted or changed.

That last line is your seatbelt. Do not trust cleaned data until you can see what changed.

I want to build this app: [DESCRIBE IT]. Tool I will use: [TOOL]. Do not give me all the code at once. Work like this: 1) Define the smallest working version. 2) Build that and let me test it. 3) After I say 'ok', add the next feature. At each step tell me exactly what to do, click by click. We are on step 1, start.

Asking for all the code at once is the most expensive beginner mistake. When it breaks you cannot find where.

Review the code or setup below for security. Give me: 1) Serious risks, most dangerous first. 2) How each could actually be exploited. 3) A concrete fix. 4) Whether secrets like passwords or keys are stored in the wrong place. 5) Things someone at my level is likely to miss. Be practical, not theoretical. Code or setup: [PASTE IT]

The hidden cost of shipping fast with AI is security. Always run this before going live.

I need to choose a tool for this job: [JOB]. I am considering [TOOL A] and [TOOL B]. My situation: [BUDGET, TECHNICAL LEVEL, TEAM SIZE]. Build a comparison table: price, learning curve, fit for my work, limitations, what happens as I grow, and whether my data is portable. At the end pick one for my situation and justify it. Mark any pricing you are unsure about as 'verify'.

Pricing goes stale fast. Always verify the numbers on the tool's own site.

I keep working on the same project with AI and I have to re explain everything every time. My project: [DESCRIBE IT]. Write me a context block to paste at the start of every new chat. Include: the project's purpose, the tools I use, my level, my preferences and rules, decisions made so far, and things I do not want done. Keep it short and dense, no filler.

This block is the single biggest efficiency jump in using AI. Write it once, use it in hundreds of chats.

You just produced this for me: [WHAT IT PRODUCED]. Now switch roles and review it as if someone else wrote it. Tell me: 1) Claims that may be wrong or unverified. 2) Important points left out. 3) Generic advice that does not fit my situation. 4) Problems I might hit if I apply this. Do not defend your own work, genuinely critique it.

This may be the most important prompt on the list. Having AI audit its own output cuts the error rate sharply.