The short answer

The best AI prompts for getting clients give ChatGPT, Claude or Gemini four things: a role (“act as a sales strategist”), real context (your offer, your past clients, the prospect's own words), numbered steps and the format you want back. You need about 15 of them to cover the whole journey: find the right people, reach them, qualify them, run the sales call and close, and follow through.

Most people try AI for client work once, type “write a LinkedIn message to get clients,” get something that sounds like a press release, and decide AI isn't much help. The tool wasn't the problem. The prompt gave it nothing to work with.

I spent years as a COO running sales and revenue teams, and the same rule applied to every new hire: the quality of the brief decides the quality of the work. AI is the same. Give it a vague request and you get a vague answer. Give it a real brief and it does in two minutes what used to take you an afternoon.

What makes an AI prompt good at landing clients?

A good client-getting prompt has four parts. Leave one out and the answer gets generic fast.

  1. A role. Tell the AI who to be: a sales strategist, a copywriter, a skeptical buyer, a financial analyst. The role sets the lens it uses.
  2. Real context. Your offer in plain words, two or three past clients, the prospect's recent post, your notes from a call. This is the part most people skip, and it's the part that makes the output usable.
  3. Numbered steps. Break the task down: first list what these clients have in common, then describe the ideal one, then tell me who to stop chasing. Steps keep the AI from jumping straight to a shallow answer.
  4. The format you want back. “Three versions under 80 words,” “a table,” “6 bullet points.” If you don't say, you'll get an essay.

One more line earns its place in almost every prompt: “Before you answer, ask me up to 3 questions if anything important is missing.” It stops the AI from guessing.

Here's the difference in practice.

Weak: Write a message to get new clients for my bookkeeping business.

Strong: You're helping me start a real conversation with a prospect, not pitch them. She runs a 12-person landscaping company and posted last week that she spends every Sunday catching up on invoices. I help trade businesses get their books current and keep them that way. Write 3 versions of a first message, each under 80 words. Mention her post, no flattery, no links, and end with one easy question she can answer in a line.

The second prompt takes 40 seconds longer to write. The message it produces is one you'd actually send.

Which AI prompts help you find the right clients?

Before you write a single message, you need to know who you're writing to. Three prompts do this work:

  • Find your best-fit client. Feed AI your three favorite past clients and have it spell out what they share: industry, size, the trigger that made them look for help, the words they use for the problem.
  • Turn your offer into one clear sentence. Have it write five versions of “I help [who] get [result] without [pain],” score each for clarity, and suggest a small fixed-price starter offer.
  • Build a list of 30 people. Ask for the specific communities, search strings and buying signals that point to people who need help right now.
1

Find your best-fit client

Prompt

You are a sales strategist who helps independent professionals choose a niche they can win in. I'm a [your role] who helps [type of client] with [result]. My three favorite past clients were: [who they were, what I did, what changed for them] x3. Step 1: List what they have in common (industry, size, stage, decision maker, the trigger that made them look for help). Step 2: Describe my ideal client in 6 bullet points, including the exact words they use to describe the problem. Step 3: Name 2 kinds of client I should stop chasing, and why. Ask me up to 3 questions first if anything important is missing.

Save those six bullets. Every prompt after this one gets better when you paste them in.

Which AI prompts write outreach that gets replies?

This is where most people start, and why most AI outreach sounds the same. Fix the voice first, then write the message.

  • Teach AI to write like you. Paste three things you wrote and have it turn them into voice rules you save in its settings.
  • Write a first message that gets a reply. One specific detail about them, one easy question, no pitch.
  • Turn your warm network into referrals. People who already know you reply far more often than strangers. (I wrote a whole guide on getting clients from people who already know you.)
  • Follow up without being annoying. Plan three or four short follow-ups, each one adding something useful instead of “just checking in.”
2

Teach AI to write like you

Prompt

Here are 3 things I've written that sound like me: [paste an email, a post and a text]. Study them and write 6 to 8 short “voice rules” covering my tone, sentence length, words I use, words I never use, and how I open and close messages. Then rewrite this using my rules: “Hi, I hope this message finds you well. I wanted to reach out to see if you might be interested in my services.” Follow my voice rules in everything you write for me from now on.

Paste the rules into your AI tool's custom instructions or project instructions, and you only do this once.

How do you use AI to qualify leads before a sales call?

Not everyone who says yes to a chat is a buyer. Two prompts protect your calendar: one plans a 15-minute fit call (what to ask, what a good and a bad answer sound like, and how to end the call either way), and one builds a short pre-call questionnaire that sits in front of your booking link. Both help you spend your real time on people with a real problem, a budget and the authority to say yes.

How can AI help you run a discovery call and close the deal?

This is the stage where AI pays for itself, and the one most freelancers never use it for. Four prompts:

  • Plan a discovery call that ends in a decision. A call outline that keeps you asking and listening, and only gets to your offer at the end.
  • Put a price tag on the problem. Turn your call notes into a conservative monthly cost of the problem, so your fee looks small next to it.
  • Rehearse the objections. Have AI play a skeptical buyer, raise the objections you dread, and score your answers.
  • Write a proposal with three options. In the client's own words, sent within a day of the call.
3

Put a price tag on the problem

Prompt

Act as a financial analyst. Here are my notes from a discovery call: [paste]. 1. List every cost of this problem you can find: hours lost (times a realistic hourly cost), missed revenue, lost leads or customers, rework, and anything they mentioned directly. 2. Show the math as a monthly and yearly figure. Use their numbers where they gave them; where you estimate, say so and stay conservative. 3. Compare the total to my price of [price]: how many weeks of the problem does my fee equal? 4. Write 2 honest sentences for my proposal that connect the cost to the investment. If you need a number I didn't capture, tell me the question to ask.

Then put that monthly figure on the first page of your proposal. A $2,000 fee reads very differently next to a $3,500-a-month problem.

Which prompts keep deals from going quiet?

Plenty of “lost” deals were never lost. Nobody followed up. The last two prompts fix that: one writes a follow-up sequence after the proposal that ends in a clear yes or no instead of months of maybe, and one runs a weekly pipeline hour, where you paste your tracker and get back who to contact, what to say and which deals to close out. A pipeline you tend every week rarely runs dry.

Does it matter whether you use ChatGPT, Claude or Gemini?

Much less than the prompt does. Every prompt in this article works in ChatGPT, Claude, Gemini and Copilot. Pick the one you'll actually open every day, and use its projects or custom instructions feature to store your ideal client and voice rules so you're not pasting them in each time. If you want to go further and have AI run some of this on a schedule, like a Monday-morning list of who to follow up with, that's what I cover in 20 AI Agents That Pay for Themselves.

Three mistakes that make AI prompts backfire

  • Sending what AI wrote without editing it. AI drafts, you decide. Cut a sentence, add one real detail only you would know, and read it out loud before you send.
  • Pasting private client data. Use public information and your own notes, and strip names and numbers a client wouldn't want shared.
  • Running prompts one at a time in new chats. Run them in order in one conversation (or one project). Each prompt builds on the last, so the tenth answer knows your ideal client and your voice.

Free download

Get all 15 prompts, word for word

The full versions of every prompt in this article, plus the fit-call plan, the discovery call outline, the objection role-play and the proposal template, as a web page and a printable PDF. Free.

Frequently asked questions

Can ChatGPT help me get clients?

Yes. ChatGPT, Claude and Gemini can define your ideal client, research where those people gather, draft personalized outreach, plan sales calls, estimate the cost of a prospect's problem and write proposals. They can't build the relationship for you. You still send the messages, show up to the calls and make the decisions.

Which AI tool is best for getting clients: ChatGPT, Claude or Gemini?

The prompt matters far more than the tool. A detailed prompt works well in ChatGPT, Claude, Gemini or Copilot. Pick the one you'll open every day, and use its projects or custom instructions feature so it remembers your ideal client and your writing voice.

How do I make AI outreach messages sound like me?

Paste three things you've written (an email, a post and a text), ask the AI to turn them into 6 to 8 voice rules, and save those rules in its custom instructions. Then edit every draft before you send it and add one real detail only you would know.

Is it okay to use AI to write messages to prospects?

Yes, as long as AI drafts and you decide. Mass-sending unedited AI messages is easy to spot and spends the trust you're trying to build. Use AI to research and draft, then personalize and send each message yourself. Never paste private client data into a public AI tool.

How many prompts do I need to get a client?

About 15 cover the whole journey: 3 to find the right people, 4 to reach them, 2 to qualify them, 4 to run the discovery call and close, and 2 to follow through. You don't need them all at once. Start with whichever stage is your bottleneck.

How long does it take to get a client using AI prompts?

It depends on your offer, price and network, so nobody can promise a timeline. What AI changes is the time each step takes. Research, drafting and call prep that used to take hours take minutes, so you can have far more real conversations each week, and conversations are what turn into clients.

Want the full system instead of the starter set? The AI Client Playbook + Kit ($29) adds 110 more prompts, 30 outreach scripts, five follow-up sequences, call scripts with 12 objection responses and a pipeline tracker. Or if you'd rather work through your pipeline together, book a free discovery call.