Where Proposal Time Actually Goes
Most consultants do not lose their hours to typing. They lose them to the blank page, to digging for the old proposal that had the right wording, and to rewriting the same background section for the fifth time this quarter. AI drafting helps with those parts. It helps much less with the thinking that wins the work, and it can hurt you if you let it do that thinking.
At Krista Mashore Coaching, we treat a proposal as a process with separate steps. You capture what the client said, decide what you will recommend, describe the scope, set the price, and package all of it in your own voice. AI is useful for some of those steps and risky for others. You save time by knowing which is which.
Watch Krista put her AI team to work. Free, live, 90 minutes.
What the Research Says, and the Warning Inside It
The best-known study on this topic is a field experiment that researchers from Harvard Business School and other universities ran with Boston Consulting Group. They published it in 2023 as the working paper "Navigating the Jagged Technological Frontier." According to that paper, 758 BCG consultants took part. On tasks that fell inside what the AI could do well, consultants using GPT-4 completed 12.2% more tasks and finished 25.1% more quickly. Their work was also rated more than 40% higher in quality than the control group's.
The same paper reports a second finding that matters more for proposals. The researchers also included a task that they deliberately placed outside the AI's abilities. On that task, consultants using AI were 19 percentage points less likely than those working without it to reach a correct answer. None of the study's tasks were proposals, so do not treat those numbers as a forecast for your own work. Pay attention to the pattern instead. AI speeds up the work it is good at. It also quietly makes people worse at the work it is bad at, because its output still sounds confident.
In a proposal, AI is good at structure, summaries, first drafts of standard sections, and tightening your language. It is bad at judging what this particular client needs, setting a fair price, and knowing what you can realistically deliver.
Build a Proposal Kit Before You Open the AI
Most of the time you save comes from preparation you only do once. Suppose you start every proposal by typing "write a consulting proposal for a manufacturing client." You will get generic text and then spend the saved time rewriting it. Give the tool your own material instead.
- Gather two or three past proposals you are proud of. Remove client names and confidential details first. These examples teach the tool your structure and tone.
- Write a service library. Describe each offer in plain language: who it serves, what you do, what the client receives, and how long it usually takes.
- Keep your standard sections ready. Save your approved wording for your background, your method, and your working terms. Paste these in as they are rather than asking AI to rewrite them each time.
- Create a short voice note. List the words you use, the words you never use, and how long your sentences usually run.
Store the kit in one place so you can paste it into a new conversation in seconds. Many AI tools also let you save instructions or reference files inside a project, which removes even that step.
Turn the Discovery Call Into the First Draft
The discovery call is the raw material for your proposal. Many consultants take notes, then sit down days later and try to remember what the client meant. You will move faster if you work from the client's own words while they are still fresh.
If you record calls, get the client's consent first and follow the recording laws that apply where you and the client are located. If you are not sure what applies, ask an attorney. Then give the AI your notes or transcript along with your kit, and ask for one output at a time:
- Ask for a summary of the client's situation in their own language, with the three problems they mentioned most often.
- Ask for a list of questions the call left unanswered.
- Ask for a draft "what we heard" section, and check it against your own memory of the conversation.
The list of open questions is often the most useful thing the tool produces. It shows you what to confirm before you commit to a scope. A short follow-up email with those questions also shows the client that you listened.
Draft Section by Section, Not All at Once
When you ask for a full proposal in one prompt, the tool fills the gaps with guesses. Drafting one section at a time keeps you in control and makes errors easier to spot. Here is a practical order to follow:
- Draft the situation and goals first. Let AI write this section from the call summary, then correct anything that sounds wrong.
- Outline the approach yourself. Write your recommendation as bullet points, then ask AI to turn them into clear paragraphs without adding any steps.
- Pull scope and deliverables from your service library. Ask AI to check that every deliverable matches a step in the approach.
- Give the tool your real availability for the timeline. Let it handle the formatting of the phases.
- Write the investment section yourself. Do not ask AI to set or suggest a price.
Once the draft is assembled, use AI one more time as an editor. Ask it to flag vague phrases, repeated points, and any sentence that promises an outcome you cannot control.
What You Must Still Write and Check Yourself
Some parts of a proposal carry real risk, and you should never trade them for speed.
- You set the price and payment terms. You know your costs, your capacity, and the value of the work, and the tool does not.
- You remove promised results. AI drafts can slip in promises of specific outcomes because so much persuasive writing contains them.
- You verify every fact. AI can invent figures, sources, and examples that sound real, so confirm each one or delete it.
- You protect client confidentiality. Read the data policy of the tool you use, and check your agreements with the client before you paste in their information. When in doubt, anonymize the details.
- An attorney reviews your contract language. Liability, intellectual property, and termination terms belong in a template that a licensed attorney has reviewed, not in an AI draft.
Run a short checklist covering these five points before you send any proposal. It protects you from the failure the BCG study described, where the work looks finished but is wrong. Your name is on the proposal, not the tool's.
Make It a System, Not a One-Off Trick
The consultants who keep their time savings are the ones who make this a routine. They keep the kit up to date, save the prompts that worked, and store discovery notes where they can find them later. A CRM helps because the call notes, the follow-up questions, and the sent proposal all live on the same client record. Our Mashore Method CRM page explains how we approach that setup.
Proposals are also just one workflow among many. Client onboarding, follow-up emails, and content can run on the same kind of kit and checklist. Our AI operating system page covers how those pieces fit together. We teach the hands-on version at our Build Your AI Dream Team event.
Your Next Step
Start small. Pull your best past proposal, write your service library, and draft your next proposal section by section with the checklist beside you. You will quickly see which steps AI handles well for you and which ones you should keep.
If you want help building a proposal workflow that fits the way you sell, contact our team and tell us where your proposals get stuck.