Straight answer: AI can cut proposal writing time from three hours to about forty minutes, but a proposal that reads like it came from a machine will lose you deals against a competitor who bothered to sound human. The fix isn't ditching AI, it's using it for structure and research, then rewriting the parts where the client needs to feel seen.
Why I started doing this in the first place
Last spring I was quoting for four projects in one week, two AI workshops and two content strategy jobs, and I did the maths on what proposal writing was costing me in hours. Each one took roughly three hours: reading the brief again, working out scope, pricing it, writing it up so it didn't sound like every other consultant's PDF. Twelve hours a week just on documents nobody had said yes to yet.
So I built a process. I fed the brief, the discovery call notes and my standard pricing tiers into ChatGPT, asked it to draft a scope and structure, then edited hard. Six months, roughly 40 proposals. It worked, mostly. Then it didn't, twice, in a way that taught me something most people writing about AI and small business skip over.
The two that went wrong
The first was a mid-sized accountancy firm in Reading, a £4,200 job for a three-month LinkedIn and email strategy. I ran the discovery call, fed the notes into the AI, got a clean draft in twelve minutes, tidied it in twenty, sent it. Lost it to a competitor charging more. When I asked the prospect why (I always ask, most people won't tell you but some will), she said the other proposal "mentioned things we'd talked about" and mine felt like it could have gone to anyone.
She was right. My draft had the right scope and the right numbers, but it had lost the specific detail she'd given me on the call, that her team hated posting on LinkedIn because a director had been trolled there in 2019 and everyone was still nervous about it. That detail mattered to her more than my pricing tiers did. The AI draft had generalised it into "addressing internal reluctance to post," which is technically accurate and completely dead.
The second was smaller, a £900 one-day workshop, but the client told me afterwards she'd had three near-identical proposal structures from three different consultants that same week. Same headers, same "problem, solution, outcomes" flow, same three pricing tiers. She said it made her feel like she was buying a template, not hiring a person. Which, in that instance, she nearly was.
What I changed, step by step
I didn't stop using AI. I changed where in the process it does the work.
- Step 1, before the call: I ask AI to draft ten questions specific to the client's industry, based on their website and LinkedIn activity, so the discovery call goes deeper than my standard question list.
- Step 2, during the call: I take notes by hand, not typed, on anything that sounds like a specific frustration, a name, a date, an internal politics detail. Those are the bits that get cut when you dump raw notes into a chatbot, because the AI treats them as noise.
- Step 3, first draft: AI builds the scope, timeline and pricing structure, the mechanical parts. This is where it earns its keep, this bit takes maybe fifteen minutes now instead of ninety.
- Step 4, the rewrite: I go through and manually reinsert every specific detail from my handwritten notes, usually two to four sentences, in the opening paragraph and the "why this approach" section. This is the part that decides whether I win or lose.
- Step 5, read it aloud: If a sentence sounds like it could be sent to any client in my industry, I cut it or rewrite it with a name, a number, or something they said.
Total time now: about 40 to 50 minutes per proposal, versus three hours before. But the win rate matters more than the time saved, and since I made this change my close rate on proposals over £1,500 has gone from roughly 1 in 4 to closer to 1 in 3. That's not a scientific study, it's one consultant's pipeline, but it's consistent enough over 25-plus proposals that I trust it.
The uncomfortable bit nobody selling AI tools wants you to hear
Here's the thing most posts about AI and proposals won't say out loud: your prospects are increasingly using AI too, on their end, to draft the brief, compare quotes, even summarise your proposal before they read it. Which means we're heading toward a world where an AI writes the brief, an AI writes the response to the brief, and then a person has to decide who to hire based on two documents that both sound reasonably competent and slightly generic. In that world, the proposal itself stops being the differentiator. What decides it is whatever happened before the document existed, the call, the follow-up email, whether you remembered the detail about the director and the trolling incident. Small business owners who think a slicker AI-written proposal is the answer are optimising the part of the sale that matters least. The document is a formality now. The relationship is the product.
That's uncomfortable if you've been sold the idea that AI will handle your sales process end to end. It won't, not the part that closes deals for services businesses under, say, £50,000 in contract value. Above that, process and documentation start to matter more. Below it, people are still hiring people.
What this looks like in practice, with real numbers
For context on where the time goes: a typical proposal for a one-day AI workshop (my most common product right now, priced between £950 and £1,800 depending on group size) breaks down like this without AI assistance: 45 minutes reading the brief and notes, 60 minutes drafting scope and outcomes, 45 minutes on pricing and terms, 30 minutes formatting and proofing. That's three hours.
With the process above: 5 minutes prepping the AI prompt with call notes, 10 minutes reviewing the draft it produces, 20 minutes rewriting the opening and the "why us" section by hand, 10 minutes on pricing checks and formatting. Roughly 45 minutes. The saving is real. The risk, if you skip the rewrite step because you're busy (and you will be tempted to, I still am some weeks), is that you send something that could have come from any AI consultant, and prospects notice more than we'd like to admit.
If you're building any kind of AI workflow into how you run a small business, whether that's proposals, content, or customer replies, the same rule applies everywhere: let the machine do the scaffolding, keep the specific human detail for yourself. I've written before about how I do this across the rest of my business without being remotely technical, which covers the tools I use day to day rather than the ones people talk about.
Where this fits with the rest of your AI content strategy
This proposal problem is really the small-scale version of a bigger issue I've run into with content generally. I once wrote about publishing 569 blog posts using AI in 45 days and having Google index only 5 percent of them, and the lesson was near identical: volume and speed from AI are useful, but the parts that get results, whether that's rankings or won deals, are the parts you can't automate away. The specific detail, the thing only you would know, is the entire value.
The same logic shows up in how my website started earning again after a long dry patch, where the recovery came from adding real specifics back into pages that had gone generic, not from producing more pages. And it's the same thing that makes a guest post pitch get accepted rather than binned, editors can tell within two lines whether a pitch was personalised or blasted to 200 sites at once.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
If your business sells anything based on a quote, a scope of work, or a proposal, this is worth testing for a month. Track wins and losses before and after you change your process, not just how much time you saved. Time saved that costs you the deal isn't a saving, it's a discount you gave a competitor for free.
If you'd rather not build this alone
I get asked fairly often by small business owners whether it's worth paying someone to set this kind of workflow up rather than fumbling through it with trial and error like I did. Honestly, if proposals or quotes are a meaningful part of how you win business and you're sending more than a handful a month, it's worth an hour with an AI consultant who can build the prompt structure and the checklist around your actual sales process, rather than guessing. It paid for itself for me within two won deals.
One more thing worth trying, if you sell anything that needs a quote based on variables (size of team, number of locations, scope of work), is building a simple interactive calculator on your site so prospects get a real number before they ever email you. I've written before about how interactive calculators convert visitors better than static pricing pages, and it pairs well with this whole approach, the calculator qualifies them, the AI drafts the scaffolding, and you spend your limited time on the two paragraphs that win the job.
Free resource: The Proposal Template That Closes.
Frequently asked questions
Should small businesses use AI to write client proposals at all?
Yes, for the structure, scope and pricing sections, where it saves genuine hours. No, for the opening paragraph and the section explaining why you're the right fit, those need specific detail from the actual conversation you had, which is exactly what clients say they notice missing when a proposal feels generic.
How much time does AI save on proposal writing?
In my own tracking across roughly 40 proposals, average time dropped from about three hours to 40 to 50 minutes per document. The saving holds up as long as you still spend 15 to 20 minutes manually rewriting the personal detail back in, skipping that step is where the risk sits.
Can clients tell when a proposal was written by AI?
Often, yes, especially when they've received three similar-looking proposals in the same week, which happens more than people admit. The tell isn't bad grammar, AI writes cleanly. It's the absence of anything that could only have come from your specific conversation with them.
What's the biggest mistake businesses make with AI in their sales process?
Assuming the document is what wins the deal. For most services businesses under roughly £50,000 in contract value, the call, the follow-up, and whether you remembered a detail that mattered to the client matter more than how polished the proposal looks. AI can write you a better looking document. It can't have the conversation for you.
Related: looking for work with me on AI? Here is how I can help.
Related reading: The AI Subscription Stack: What I Pay For (and What I Cancelled) in 2026 and I Got AI to Chase My Late Invoices. Here's What Happened.
