You halve proposal time by drafting less from scratch, not by rushing the pricing. Build a reusable proposal kit once (winning past proposals, a library of standard sections, tone notes) inside a ChatGPT or Claude project, then give it each client's call notes and let it draft the concept, schedule and inclusions. You set the prices and check every claim.
The time you save is almost all in the writing. Pricing, supplier availability and the final read-through take as long as they did before, and they should. Below is where the hours go, what to put in the kit, four prompts in the order to run them, and a worked example with timings.
Where the hours in an event proposal actually go
Before changing anything, time your next proposal stage by stage. Most planners find the writing is spread across tasks they'd never think of as writing. Here's an illustrative breakdown for a mid-sized corporate event of 80 to 150 guests, using my estimates for a planner who works from a template:
| Stage | Before (minutes) | With the kit (minutes) | What changes |
|---|---|---|---|
| Rereading the enquiry and call notes | 20 | 10 | AI summarises and lists the gaps |
| Concept and theme | 60 | 20 | AI drafts two directions; you choose and adjust |
| Outline schedule | 30 | 10 | Drafted from the brief's fixed times |
| Inclusions and supplier options | 45 | 25 | AI writes the descriptions; you still confirm availability |
| Pricing | 45 | 45 | Unchanged: your sheet, your margins |
| Standard sections (about us, process, terms) | 40 | 5 | Copied from your section library |
| Formatting and images | 50 | 20 | The template carries the layout |
| Checking and sign-off | 30 | 30 | Unchanged, on purpose |
| Total | 320 (5h 20m) | 165 (2h 45m) |
Two rows don't move at all. If someone promises you a proposal method that shortens pricing and checking, they're describing a way to send out mistakes faster.
Build the proposal kit once (allow an afternoon)
The kit is what makes the drafts sound like you instead of like every other planner using the same assistant. Gather these into one folder:
- Three proposals that won, plus one that lost, with a line on why it lost if you know. Remove client names and prices from the copies, or replace them with placeholders.
- A section library. About you, how you work, what your fee includes and excludes, the payment schedule, cancellation, insurance, what happens if numbers change. You write and approve these once; the AI copies them word for word rather than rewriting them each time.
- Tone notes, five to eight lines. For example: "We say guests, not attendees. No exclamation marks. Short paragraphs. We describe what guests will notice, not what we'll 'curate'."
- A supplier roster. Who you use, what each can deliver, and what you must never promise. "Our florist can't do more than 20 centrepieces on a same-day turnaround" is the kind of line that stops the AI proposing something you can't deliver.
- A banned-claims list. "Award-winning" unless you have the award, "guaranteed", capacity figures you haven't checked, any mention of past clients who haven't agreed to be named.
- The proposal template in Word, Google Docs or your design tool, with headings in a fixed order.
Then set up a project. In ChatGPT, Projects keep chats, files and instructions together, and they're available on the free plan as well as paid ones. In Claude, free users can create a small number of projects, but project instructions (the standing rules) need a paid plan. Upload the kit as project files and paste the tone notes, roster rules and banned claims into the instructions. If you handle several clients at once, keeping client work separate in ChatGPT Projects explains how to stop one client's details bleeding into another's draft.
Capture the brief so the AI has something real to work from
Vague notes produce vague proposals. After every discovery call, spend five minutes filling this in:
CLIENT: company, contact, role, who signs off
PURPOSE: why this event exists, in the client's own words
DATE: preferred date, flexibility, day of week
GUESTS: number, who they are, what they expect
FORMAT: seated / standing / mixed; timings that are fixed
MUST HAPPEN: ...
MUST NOT HAPPEN: ...
BUDGET: stated figure, what it covers, how firm it sounded
LAST TIME: what they liked or disliked about previous events
DECISION: when they need the proposal, how they'll choose,
who else they're talking to (if known)
Filled in after a 25-minute call with an illustrative software company planning its summer staff party, it might read:
CLIENT: software company, 140 staff; contact is the office manager;
the COO signs off
PURPOSE: "People only see each other on video calls. We want the
teams to actually talk to each other."
DATE: a Thursday in late June or early July; avoid quarter-end week
GUESTS: 120 expected, ages 22-60, several with young families
FORMAT: mixed, mostly standing; finish by 11pm (last trains)
MUST HAPPEN: short thank-you from the CEO, under 10 minutes
MUST NOT HAPPEN: "No forced team-building. No one wants a trust exercise."
BUDGET: about $140 a head, "could stretch a bit for the right venue"
LAST TIME: rooftop bar, "ended at 9 because everyone was just standing
around"; food ran out by 8
DECISION: proposal by the 14th; also talking to one other planner
The two most useful lines are in the client's own words: the purpose and last year's complaint. They are what Prompt 4 will later pull into the proposal, so write them down as said, not paraphrased.
If you record the call with the client's agreement, a transcript works too, but boil it down into this template first. Raw transcripts are long and the assistant weights small talk as heavily as the brief.
The drafting sequence: four prompts, in this order
Running these as separate steps keeps you in control of the decisions. One giant prompt asking for a finished proposal gives you a finished-looking document built on assumptions you never saw.
Prompt 1: the gap check
Here are my call notes for a new event enquiry. Using the kit in
this project:
1. Summarise the client's goal in one sentence, in their language.
2. List what's missing that I need before I can price this (max 6).
3. List the assumptions you'd make if I don't get answers.
For the staff party above, the reply might look like this (illustrative):
Goal: an evening where staff from different teams actually talk to each other, without anything that feels like forced team-building. Missing: 1. Is $140 a head inclusive of drinks, or food only? 2. Are partners or children invited? 3. Any dietary or accessibility needs known now? 4. Does the CEO need a microphone and screen? 5. Is the company paying for travel home after 11pm? Assumptions if unanswered: seated dinner for 120; drinks on a bar tab capped at the budget; staff only.
Most of that is good. The one to catch is the first assumption. The notes said "mostly standing", and a seated dinner for 120 changes the venue shortlist and the price per head more than anything else in the brief. The assistant slid towards the most common corporate format. Correct it in the chat ("standing, with some seating; don't assume a dinner") before Prompt 2, or the concepts will be built on it.
Send the missing items to the client as a short email the same day. It moves the proposal forward and shows you were listening.
Prompt 2: two concept directions
Suggest two distinct concept directions for this event. For each:
a working name, a two-sentence description, how it serves the
client's stated purpose, and three details guests would notice.
Use only capabilities in the supplier roster. Flag anything that
would need a supplier we don't have.
Prompt 3: the draft
Draft the proposal for concept [A/B] using the template headings,
with these changes: [your notes].
Copy the About us, How we work, Payment and Cancellation sections
word for word from the section library.
Leave the Investment section as the placeholder [PRICING TABLE].
Follow the tone notes. Under 1,200 words, excluding library sections.
Prompt 4: the client-specific pass and claim list
Reread the call notes. Find three places where the draft could use
the client's own words or refer to something they told us (a past
event, a worry, a guest). Suggest the edits.
Then list every factual claim in the draft (dates, days of the week,
capacities, supplier promises, numbers) as a checklist.
The client-specific pass is where a draft stops reading like every other proposal. Take a before and after from the same staff party. Before, from Prompt 3: "Guests will enjoy a curated evening of fine food and entertainment in a stunning setting." After, with Prompt 4's suggestion: "You said last year's party ended at 9 because everyone was just standing around. This time food arrives in three waves from 7.30pm, so nobody's hungry by 8, and a two-round music quiz at 8.45pm gives people from different teams a reason to sit together, without anyone being made to." The second paragraph answers the client's complaint, uses their words and keeps the "no forced team-building" line honest.
The checklist from Prompt 4 becomes the starting point for your red-pen pass. A slice of it, for the same party:
1. Thursday 2 July (date and weekday)
2. Venue holds 150 standing, 90 seated
3. Terrace available until 10.30pm
4. Florist: 12 high-table arrangements
5. Food in three waves from 7.30pm
6. Last entry to bar 10.45pm
Item 3 is the one to chase. Neither the brief nor the venue's details mention a terrace; it came from a past winning proposal in the kit, for a different venue that had one. That's the price of giving the assistant real examples, and the claim list is where it surfaces.
Keep pricing out of the chat window
Asked to price an event, an assistant will produce confident, plausible numbers that have nothing to do with your supplier deals or margins, and it may add them up wrongly. Price in your own spreadsheet, where the formulas are visible; the method in building event budgets and run sheets with AI gives a layout that separates quotes from estimates.
Once the prices are final, paste only the figures you want the client to see and ask the assistant to format the investment table and write a two-line explanation of each option. If you offer tiered options, the AI is useful for suggesting what belongs in each tier ("what would a guest notice if we dropped from the middle option to the lower one?"). Deciding what each tier costs stays with you. Afterwards, check the formatted table against your sheet to the cent.
Asked that tier question about the staff party, an assistant might say: "Dropping to the lower option, guests would notice two food waves instead of three, a playlist instead of the live duo during arrivals, and a two-hour bar tab instead of three. They probably wouldn't notice the change from printed menus to a single board, or fewer flowers on the high tables." That's a useful list for building the tiers, with one fix. The third food wave is what answers last year's complaint, so it moves back into every tier and the saving comes from somewhere else. The assistant ranks by what's typical; you rank by what this client told you.
The red-pen pass: 30 minutes you don't skip
- Every name spelled correctly: client, company, venue, speakers.
- Date and day of the week agree. Assistants regularly get the weekday wrong for future dates, so check against a calendar.
- Venue capacity fits the layout you're proposing; standing capacity is often far higher than seated.
- Every supplier promise is on the roster and the supplier is free on the date.
- Nothing from another client's proposal survived: names, themes, a guest count from a different event. This is the price of using past proposals as examples.
- Library sections are unchanged. Compare a paragraph or two with the source files. The failure looks harmless: the library says "The booking deposit is non-refundable", and the draft, asked to copy it word for word, comes back with "The booking deposit is refundable up to 60 days before the event", borrowed from a softer line in one of the winning examples. One word changes your cancellation terms.
- The inclusions list matches the investment table line by line.
- The validity date and any date-hold terms are present.
The proposal mistakes that cost you the job covers the wider list.
Worked example: a car dealership's new-model launch
This example is illustrative. A car dealership asks a planner for an evening launch in its showroom for 80 invited customers, with a budget of about $25,000. The planner's last comparable proposal, written from scratch, took about five and a half hours.
The gap check took eight minutes and produced four questions for the dealership: could cars be moved out of the showroom for the evening, was there power for a DJ and lighting at the far end, did the budget include preparing the vehicles, and who would sign off. The dealership answered three the next morning.
Prompt 2 offered an "after hours lounge" concept, with the new model revealed from under a cover mid-evening, and a "test drive social" with short supervised drives. The second concept had a problem the model didn't flag: it paired the drives with a cocktail bar. The planner kept the idea but made the bar alcohol-free until the last drive slot ended, and wrote that into the schedule. It's a good reminder that AI suggests what sounds appealing, not what's safe or licensable.
Timings for the rest: 20 minutes adjusting the draft from Prompt 3, 25 minutes on inclusions and supplier checks, 45 minutes on pricing, five minutes for library sections, 22 minutes on layout and images, and 35 minutes on the red-pen pass, which caught a Friday date labelled as a Thursday. The total came to roughly two hours 55 minutes. Whether it wins depends on the concept and the price; the method only means the planner has the rest of the afternoon back.
The revision round: changes without a rewrite
Most event proposals come back at least once. The client wants 20 more guests, a later finish, the other concept's entertainment with this concept's food. This is where planners lose the time they saved, because they reopen the document and start editing by hand, and the changes ripple into sections they forget to update.
Instead, go back to the same project chat and give the assistant the change list and the current version together:
The client has asked for these changes: [list].
Here is the current proposal: [paste or attach].
1. List every section affected by each change, including ones
that mention guest numbers, timings or inclusions indirectly.
2. Redraft only those sections. Mark each changed paragraph [REVISED].
3. List anything in the change request that affects price, so I can
update the pricing sheet before the proposal goes back.
Step 1 is the valuable one. Twenty extra guests touch the capacity line, the catering inclusions, the staffing note, the cloakroom and possibly the venue layout; a person editing in a hurry fixes two of the five. Ask for "20 more guests and a finish at midnight instead of 11pm" on the staff party, and step 1 might come back with seven sections: the capacity line, the food quantities in all three waves, bar staffing, the cloakroom, the schedule, the quiz timing (now too early in a longer evening) and travel home, since the brief said 11pm was set by the last trains. The last one is the kind of knock-on nobody spots while editing by hand, and it may mean adding taxis to the price. Step 3 keeps the revised price coming from your sheet rather than from a guess, and the [REVISED] markers let the client see what moved without rereading everything. Save each version with a date in the file name, and never let the assistant's copy become the only copy: the approved document lives in your own drive.
A small habit helps here too. When a proposal wins, add it to the kit's winning examples (with names and prices removed) and delete the oldest one. The kit stays current with what your clients actually buy, and the drafts drift towards your best recent work rather than something from three years ago.
When dedicated proposal software is worth paying for
A general assistant plus a good template covers most small planners. Proposal software starts to earn its fee when you send enough proposals that tracking matters: knowing when a client opened it, letting them pick options and sign online, and keeping versions in one place. If you're weighing that up, AI proposal software versus a general AI assistant compares the two on cost and effort. Either way, the kit you build here moves with you, because it's your content, not the tool's.
Further reads
- How Consultants Use AI to Write Proposals in Under an Hour — A tighter version of the method for short, repeatable proposals.
- How to Set Up Claude Projects as a Shared Team Assistant — Share the proposal kit with a colleague so both of you draft the same way.
- How to Build a Brand Voice Guide That AI Can Follow — Turn your tone notes into a fuller voice guide.
- How to Automate Sales Follow-Ups With AI Without Being Pushy — What to send when the proposal goes quiet.
- How to Mention AI Use in Client Contracts and Proposals — Whether and how to mention AI use in your terms.
- Best AI Tools for Small Event Planning Businesses — The wider tool set for a small events business.
- A Wedding Planner's AI Workflow From Enquiry to Final Timeline — One illustrative planner's year with AI, stage by stage: consultation notes, proposal, suppliers, guests, confirmations and the wedding-day timeline.
- How to Write Wedding Flower Proposals With AI — A florist's method for AI-assisted wedding proposals: consultation brief, recipe costing, prompts for the words, and the seasonality check AI can't do.
- How Caterers Use AI to Quote Events and Plan Menus — How a small caterer can use AI to go from messy enquiry to costed quote and proposal, with a worked 70-guest quote and prompts to copy.
- Which Event Planning Tasks Should You Hand to AI First? — Twelve event planning tasks scored for AI, the five to hand over first with worked examples, and the ones to keep even though AI will attempt them.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: OpenAI help article on Projects in ChatGPT; Claude help article on creating and managing projects (both checked September 2026).