How I Use AI to Write Client Proposals in Under 5 Minutes

For two years, my bottleneck wasn't leads. It was proposals.

A great discovery call on Tuesday would sit in my head as "the Martinez proposal I need to write" until Friday, when I'd finally open a blank doc, stare at it, copy-paste from a prior proposal, manually rewrite every section, second-guess the pricing, and send at 9pm — four days after the call.

By Friday the prospect had cooled. By the next Tuesday they'd "gone with someone else."

Slow proposals aren't just inconvenient. They're a conversion killer. The fix wasn't discipline. It was a workflow that got drafts out of my head and in front of the client before dinner on the day of the call.

The old way vs. the AI way

Old way: open a blank doc, find a similar past proposal, copy/paste, rewrite intro, rewrite scope, rewrite price table, re-check pricing math, rewrite close, send 2-4 days later.

AI way: I keep a structured notes template open during the discovery call. The moment we hang up, I paste those notes into a saved master prompt, get a 90%-there draft in 30 seconds, spend 3-4 minutes personalizing the open, the "what I heard" section, and the close, then hit send.

The proposal lands in the prospect's inbox while the call is still warm in their mind. That timing alone has moved my close rate from ~30% to ~45%.

The three-part workflow

There are only three moving parts. Once they're set up, you reuse them on every proposal.

Part 1: A structured call notes template

During every discovery call, I fill in this template in Notion (or Google Docs — doesn't matter):

CLIENT
- Name:
- Business:
- Role / title:
- What does their business actually do (in my words, not theirs):

THE PROBLEM
- What they said the problem is:
- What I think the underlying problem actually is:
- What's it costing them (time, money, stress, missed opportunity):

THE GOAL
- What they said "success" looks like:
- Timeline pressure (is there a real deadline or is it soft):

SCOPE SIGNALS
- Anything they mentioned wanting that's NOT in my usual package:
- Anything they explicitly said they DON'T need:

BUDGET
- Price range mentioned (or my read on it):
- Any phrases they used like "we don't have much budget" or "money isn't the issue":

WORKING STYLE
- Decisions made by one person or a committee:
- Communication preference:
- Red flags or "hmm" moments on the call:

This template does 80% of the work. The quality of your notes determines the quality of the proposal — not the prompt.

Part 2: The master proposal prompt

I have a single prompt saved that I reuse for every proposal. It lives in my password manager (honestly) because it's that important to me.

You are writing a client proposal for me.

ABOUT ME:
- I'm a [your profession] working with [your niche].
- My standard packages are:
  - [Package A — name, what's included, price, timeline]
  - [Package B — name, what's included, price, timeline]
  - [Package C — name, what's included, price, timeline]
- My tone: [3-5 adjectives — e.g., direct, warm, specific, not corporate].
- Things I never say in proposals: "leverage," "unlock," "synergize," "elevate."

ABOUT THIS CLIENT:
[paste your filled-in notes template here]

TASK:
Write a proposal for this client that includes:
1. A one-sentence opener that references something specific they said (use the "What they said the problem is" field).
2. A short "What I heard" section (2-3 sentences) that plays back their situation in my voice.
3. A recommendation paragraph: which package I'm proposing and why it fits.
4. A deliverables section: bulleted, specific, outcome-framed (not just task lists).
5. A timeline with 2-3 phases.
6. An investment section: the price, and one line on what it includes.
7. A "What I'll need from you" section: 2-4 items, specific.
8. A closing paragraph that makes the next step frictionless.

STYLE CONSTRAINTS:
- Under 600 words total.
- Short paragraphs, no more than 3 sentences each.
- No bullet list is longer than 5 items.
- Confident, specific, and referential to their situation. Avoid flattery.

That's the whole prompt. Copy it, fill in the [your profession] sections once, and never write one from scratch again.

Part 3: The personalization pass

The AI draft is 90% done. The last 10% is what separates proposals that close from proposals that look like AI.

My personalization checklist:

• The opener. Replace the AI's opener with a single sentence that references the most specific thing they said on the call. "That bit about losing a whole Tuesday to invoicing stuck with me" beats any AI intro.
• The "what I heard" section. Read it and ask: does this sound like me paraphrasing them? If either half is off, rewrite 1-2 sentences.
• One past-project reference. Drop in a single line like "This is the third time I've seen this specific pattern in [their industry] — same fix worked both prior times." Instant credibility.
• The close. AI closes tend to be slightly formal. Rewrite to your natural send-off.

That's four small edits. Three to four minutes. Done.

A worked example (real, lightly anonymized)

Call notes excerpt:

• Client: Jess, founder of a 6-person branding agency.
• Problem: manual onboarding eating 3+ hours per new client, team members sending inconsistent welcome docs.
• Goal: "I want a new client to feel like they've joined a real agency, not a group chat."
• Budget signal: "We've got budget for this, just don't want to over-engineer it."
• Package fit: Onboarding Sprint — $4,800 / 2 weeks.

AI draft opener (generic): "Thanks for taking the time to share what you're working on, Jess — I enjoyed our conversation."

My rewrite: "Jess — 'a real agency, not a group chat' is going to live rent-free in my brain this week. That's exactly the gap we'd be closing."

Three seconds of edit. Twenty points of warmth.

Why the speed is the leverage

Here's the part most people miss when they hear "AI writes your proposals."

The win isn't that AI writes better proposals than you. It doesn't. A careful, handwritten proposal from a practitioner who lived the client's problem will always beat an AI draft.

The win is that you now write proposals on the same day as the call. That single change — measured in hours saved, not words saved — is what drives the close rate up.

My proposal stats since moving to this workflow, measured over 100+ calls:

• Proposals sent same-day: went from 15% to 90%.
• Average time from call to proposal: went from 3.4 days to 4 hours.
• Close rate on proposals: went from ~30% to ~45%.
• Proposals I actively avoided writing: went from 4-5 per month to ~zero.

That last metric is the quietly important one. The proposals you never write because you're dreading them are the ones you never close.

What AI still can't do

I'll save you the Luddite fear and say what's actually true: AI struggles with three things in proposals.

1. Pricing judgment. It'll happily propose $3,500 for a job that should be $8,500. Don't let it near your pricing math.
2. Strategic recommendations. It can write good language around a strategy; it can't tell you the right strategy. That's your job.
3. Reading the room. If your prospect is nervous, price-sensitive, or flag-waving a specific objection, AI misses it unless you explicitly name it in the notes.

Keep those three out of AI's hands and the rest is safe to delegate.

Frequently asked

Should I disclose that I used AI? If the prospect asks, yes. Unprompted, no. AI-assisted proposals are becoming as normal as spell-checked ones. The relevant disclosure is your judgment and editing, not the drafting tool.

What's the best tool for this? I use Claude for proposals because its tone lands closer to mine on the first try. GPT-4 is a close second. The right answer is whichever model you'll actually open same-day.

Won't every proposal start sounding the same? Only if you skip the personalization step. The opener and the "what I heard" section are the bits the prospect reads most carefully, and those are the ones you rewrite by hand.

If proposals are your bottleneck, start by building the notes template (Part 1) this week. It'll do more for your close rate than any prompt. Once that's a habit, the AI step is a 20-minute setup. Pair it with my three-email discovery call follow-up sequence and your entire post-call workflow is done in well under 30 minutes.