Which AI Model to Use for Which Job? A Practical Guide
Not sure which AI model to use? Match the model to the task: fast small models, strong reasoning models, or local models, with simple rules of thumb.
How to write AI prompts that get useful answers: 10 practical tips for everyday chats, each with a before-and-after example you can copy and adapt.
To write better AI prompts, tell the assistant who the answer is for, what you want it to produce, and what a good result looks like. Most weak answers come from prompts that leave those three things to guesswork. The ten tips below fix that, each with a before-and-after example you can copy.
These tips are for everyday chats: emails, explanations, plans, and summaries. If you want to hand a multi-step job to an agent, the rules shift a bit, and we cover that in prompting AI agents vs prompting chatbots.
An AI assistant has no idea who you are, what you already know, or why you’re asking. It fills every gap with the most generic assumption available. That’s why short prompts tend to produce bland, middle-of-the-road answers.
A good prompt isn’t longer for the sake of it. It simply removes the guesses that matter.
Name the output: an email, a list, a table, a two-paragraph explanation. “Tell me about” invites an essay.
Before:
Tell me about meal prepping.
After:
Give me a 5-day meal prep plan for one person: lunches only,
under 30 minutes of cooking on Sunday, as a shopping list
plus a short prep schedule.
The same topic needs a different answer for a child, a beginner, and an expert. Tell the AI who will read the result.
Before:
Explain how vaccines work.
After:
Explain how vaccines work to a curious 10-year-old, in about
150 words, using one everyday comparison.
The assistant can’t see your situation. One or two sentences of background often change the whole answer.
Before:
Write a message asking for a deadline extension.
After:
Write a short message to my manager asking to move the report
deadline from Friday to Tuesday. Reason: the sales data arrived
three days late. Our tone at work is friendly but direct.
If you don’t set a length, you get whatever the model thinks is typical. Be explicit: word count, number of bullets, or “one sentence.”
Before:
Summarize this article.
After:
Summarize this article in 3 bullet points, each under 20 words,
then one sentence on what it means for a small online shop.
Describing a tone is hard. Showing one is easy. Paste a sample and ask the AI to match it.
Before:
Write product descriptions for my candles. Make them sound nice.
After:
Write product descriptions for these three candles. Match the
style of this example:
"Cedar & Smoke. Like a cabin weekend, minus the drive. 40 hours
of slow, even burn."
Keep each under 25 words.
A huge request in one message gets a shallow answer to everything. Ask for an outline first, then expand the parts you care about.
Before:
Write my whole business plan for a dog-walking service.
After:
Let's build a business plan for a dog-walking service step by step.
First, list the 6 sections it should have, with one line on what
each covers. Don't write the sections yet.
When you’re not sure what details matter, let the assistant interview you first. This is one of the most underused tricks.
Before:
Help me plan a birthday party.
After:
I want to plan a birthday party for my partner. Before suggesting
anything, ask me up to 5 questions about what you need to know.
Constraints are as useful as instructions. If you hate corporate jargon, exclamation marks, or long intros, say so.
Before:
Write a LinkedIn post about my new job.
After:
Write a LinkedIn post announcing my new job as a data analyst at a
nonprofit. No buzzwords, no emojis, no "I'm thrilled to announce."
Under 80 words.
For anything creative or subjective, one answer is a coin flip. Ask for several and pick.
Before:
Give me a name for my bakery.
After:
Suggest 10 names for a small neighborhood bakery known for
sourdough. Mix playful and simple ones. Add a one-line reason
for each.
The first answer is a draft. Tell the assistant exactly what to change rather than rewriting your prompt from scratch.
Before:
(new chat) Write a shorter, friendlier version of a cover letter...
After:
Good start. Make paragraph two shorter, drop the sentence about
teamwork, and make the opening sound less formal.
Before you hit enter, check your prompt against this table.
| Question | If the answer is “no,” add… |
|---|---|
| Does it name the output? | “Write a…”, “List…”, “Make a table of…” |
| Does it say who it’s for? | The reader and their level |
| Does it include my situation? | One or two lines of background |
| Does it set length or format? | Word count, bullets, sections |
| Does it rule out what I don’t want? | Tone, words, or topics to skip |
You won’t need all five every time. A quick question needs none of them. A message you’ll send to your boss probably needs most.
When in doubt, fill in this pattern:
I need [output] for [audience].
Context: [one or two sentences about your situation].
Make it [length/format]. Avoid [things you don't want].
If anything is unclear, ask me before you start.
It works for emails, summaries, study notes, travel plans, and most other daily tasks.
Even a great prompt can produce a confident mistake. AI assistants sometimes invent facts, numbers, or sources that sound right. For anything that matters, such as figures, legal or medical details, or quotes, verify the answer against a reliable source before you use it.
It also helps to keep sensitive details out of prompts. You rarely need real names, account numbers, or private documents to get a useful answer. Our guide on sharing your data with AI safely covers what to leave out.
Once you start asking an AI to do things, like edit files or build a small website, a prompt becomes a brief: a goal, context, constraints, and a finish line. That’s the territory of AI agents. If you’re curious about the difference, start with what an AI agent is. Tools like BotQuarters go a step further and have a team lead agent write those briefs for a whole team of coding agents.
State the output you want, who it’s for, and the context only you know, then set a length or format. Add what to avoid if you have strong preferences. Treat the first answer as a draft and refine it with specific follow-up requests.
Not automatically. A prompt should be long enough to remove the guesses that matter and no longer. A clear two-sentence prompt with the right context often beats a rambling paragraph.
Politeness doesn’t hurt, but clarity matters far more. “Please” won’t fix a vague request, while a specific audience, format, and goal usually will. Write the way you’d brief a helpful colleague.
Long prompts with many instructions can bury the one that matters. Put your most important rule near the top, keep requests focused, and split big jobs into steps. If it still misses something, point to it directly in your follow-up.
Not sure which AI model to use? Match the model to the task: fast small models, strong reasoning models, or local models, with simple rules of thumb.
AI productivity tips for real life: everyday writing, summarizing, planning, spreadsheet, and research tasks you can hand off to AI today, with examples.
Learn to code with AI the right way: use it as a tutor, not an answer machine. Practical prompts, a weekly routine, and habits that build real skill.