7 min read
How many proposals have you lost not because your price was wrong, but because you sent it four days after the call, when the client had already moved on?
For most solo service businesses, the proposal is where momentum goes to die. The discovery call goes well. The client is excited. Then you need a free afternoon to turn your scribbled notes into a polished document, and free afternoons are rare. By the time it lands in their inbox, the excitement has cooled and a competitor has replied.
This guide shows you a repeatable, 20-minute workflow for turning a discovery call into a client-ready proposal with AI. It works with ChatGPT, Claude, or Gemini. You do not need any paid add-ons, and you stay in charge of the two things AI should never decide for you: your price and your promises.
Before your next call: build the skeleton once
AI writes much better proposals when it fills in your structure instead of inventing one. Spend thirty minutes, one time, creating a proposal skeleton. Open your best past proposal and reduce it to its sections. A good skeleton for a service business looks like this:
- The situation: the client’s problem in their own words.
- The goal: what success looks like, ideally with a number or a date.
- The approach: your process in three to five steps.
- Deliverables and timeline: exactly what they get and when.
- Investment: your price options (you fill this in, always).
- What’s not included: the scope boundary that prevents arguments later.
- Next step: one clear action, such as “reply yes” or “book a kickoff.”
Save this skeleton, plus one or two of your strongest past proposals, as reference material. In ChatGPT or Claude, the easiest home is a Project, a workspace with saved instructions and files that every new chat inside it can see. In Gemini, you can store it as a reusable skill. Either way, you set it up once and reuse it for every proposal.
During the call: capture the client’s exact words
The single biggest upgrade to an AI-written proposal is a real record of what the client said. Vague notes produce vague proposals. You have three options, from easiest to most private:
- Your meeting app’s built-in transcript. Zoom, Google Meet, and Teams all offer transcripts on many plans.
- A local notetaker. If the conversation is confidential, Google’s free experimental AI Edge Foresight app transcribes on an Apple Silicon Mac without sending audio to the cloud.
- Typed notes plus a voice memo. Right after the call, record two minutes summarizing what you heard while it is fresh. This is the same capture habit we described in our guide to turning voice memos into searchable notes.
Whatever you choose, ask permission to record or transcribe at the start of the call. It is both polite and, in many states, legally required. And during the call, ask the questions that make proposals easy: What happens if you do nothing? What have you tried? Who else needs to approve this? What does your budget range look like? When do you need results?
Minute 1 to 5: extract the brief
Do not ask AI to write the proposal yet. First, have it pull out the facts. Paste your transcript or notes and use a prompt like this:
“From this discovery call, extract: (1) the client’s main problem in their own words, with 2 or 3 direct quotes, (2) their goal and any numbers or deadlines, (3) constraints such as budget signals, timing, or tools they already use, (4) who makes the decision, (5) anything they said they do not want. Use only what is in the notes. If something is missing, list it under ‘Unknowns’.”
Read the brief carefully. This is your moment to catch misunderstandings. If the “Unknowns” list includes something important, like the budget, send a quick follow-up email before you write anything. A one-line question now beats a rewritten proposal later.
Minute 5 to 10: draft against your skeleton
Now ask for the draft:
“Using the brief above and my proposal skeleton, write a proposal for this client. Open with their problem using one of their exact quotes. Keep it under two pages. Match the tone of my sample proposals. Leave the Investment section as [PRICE OPTIONS] for me to fill in. Do not promise results, timelines, or deliverables that are not supported by the brief.”
Two details in that prompt do a lot of work. Quoting the client back to themselves makes the proposal feel personal and proves you listened. Leaving the price blank keeps the most important decision in your hands.
Minute 10 to 15: set the price yourself
AI does not know your costs, your capacity, or what this client is worth to your business. Let it help you think, but not decide. A useful middle ground is to ask it for options rather than numbers:
“Suggest three package tiers for this project, a lean option, a recommended option, and a premium option, describing what each includes. Do not include prices.”
Then you price each tier based on your own rates. Three options usually work better than one because the client is choosing which way to work with you, not whether to. If you quote jobs from photos or site details, our story of a house painter producing same-hour quotes shows the same principle applied to trades.
Minute 15 to 20: have AI argue with you
This is the step most people skip, and it is where AI is most valuable. Ask it to play the client:
“Read this proposal as a skeptical version of this client. What questions would you have? What sounds vague, risky, or too expensive? What objection is most likely to stop you from saying yes?”
Fix the two or three best points it raises. Typical catches include a deliverable that sounds open-ended, a missing date, or a section full of your industry jargon. Then do a final read yourself, out loud, checking every name, number, and date against the brief.
After you hit send: the follow-up plan
A fast proposal still needs a follow-up. Before closing the chat, ask the AI to draft two short follow-up emails: one for three days later (“happy to walk through any questions”) and one for a week later that offers a specific next step. Save them as drafts or in your CRM.
If you track leads in a CRM, the AI notetaker and prospecting features in tools like HubSpot Starter can handle some of this capture and follow-up automatically. And if your business runs on Google’s apps, Google’s new Gemini small business hub connects to Workspace so your notes, drafts, and calendar live in one place.
Mistakes that make AI proposals backfire
- Sending the first draft. Clients can feel generic text. Your edits are what make it yours.
- Letting AI pad the scope. AI loves to add “bonus” deliverables. Every extra line is something you will be asked to deliver.
- Pasting sensitive details without thinking. Strip account numbers, health information, and anything under NDA before you paste, or use a local tool for those calls.
- Skipping the “not included” section. Scope boundaries are the cheapest insurance a freelancer can buy.
Your first proposal sprint
Here is your homework. Build the skeleton this week. On your next discovery call, set a timer for 20 minutes immediately afterward and run the five steps: brief, draft, price, challenge, send. Even if it takes 40 minutes the first time, you will have sent a tailored proposal the same day, which already puts you ahead of most of your competition.
What is the longest you have ever taken to send a proposal, and what did it cost you?
Related reading
- How a Solo House Painter Turned Customer Photos Into Same-Hour Quotes With ChatGPT (An Illustrative Story)
- HubSpot Starter Now Has an AI Notetaker and Prospecting Agent. Is It Finally Enough CRM for a Solo Seller?
- Google Gemini Just Got a Free Small Business Hub: What a Solo Owner Should Set Up First
- How to Turn a Shoebox of Receipts and a Phone Full of Voice Memos Into Searchable Notes With ChatGPT
