Live class resource

Turn a Discovery Call into a Proposal

Massachusetts AI Coalition. Boston. September 17, 2026.

Everything from tonight is on this page.

Reed Jones, The AI Department

Start here

Working along with me tonight? Do this.

  1. 1Copy Prompt 2 below.
  2. 2Grab the sample transcript, or use your own call.
  3. 3Paste both into Claude, ChatGPT, or Gemini.

If you did not bring a call of your own, use the sample.

What you are looking for in a transcript

Every prompt on this page is built around the same filter. If a detail does not map to productivity, reputation, or profitability, it is either a constraint or it is noise.

Pain.

How they describe the problem, in their words.

Outcome.

What fixed looks like to them.

Constraints.

Budget, timing, policy, past bad experiences.

Trust.

Who decides, and what they volunteered without being asked.

Gaps.

What a proposal needs that the call never gave you.

Everything maps to productivity, reputation, or profitability. If it maps to none of them, it is a constraint or it is noise.

Before you record anything

  1. Ask. Say what it is for. “I record these so I can write you an accurate proposal instead of a vague one.”
  2. Know where the transcript goes. Which tool, whose servers, what happens to it afterward.
  3. Take out what the tool does not need. Health information, account numbers, anything about a third party who was not on the call.
  4. Use approved tools and follow governance. Your employer's policy, your client's requirements, and any rules that cover your industry beat your preference.
  5. You sign it, so you verify it. Every price, date, scope line, and promise gets human eyes.

The main event

The four prompts

Prompt 1

Build the question list, before the call

Use this the day before. You cannot extract what was never said.

You are helping me prepare for a discovery call.

CONTEXT
Who I am and what I sell: [ONE OR TWO SENTENCES]
Who I am meeting and what I know so far: [COMPANY, ROLE, HOW THEY FOUND YOU]
What I think they might need: [YOUR BEST GUESS, OR "UNSURE"]

TASK
Write me a discovery call question list designed so the recording of this call contains everything I need to write a proposal afterward.

The call must produce all six of these. Group the questions accordingly:
1. The problem in their own words
2. What it is costing them, in their number: money, hours, deals, or turnover
3. What "fixed" looks like to them, specifically
4. Who decides, and what else has to be true for a yes
5. Constraints: budget, timing, policy, and any past bad experience with this
6. What is explicitly not in scope

RULES
- Write open questions. No question a person can answer with yes or no.
- For each of the six, give me one main question and one follow-up that pushes for a specific number or example.
- Flag the two questions people most often skip, and say why they matter.
- No more than 12 questions total. I have to actually ask these.

The one to never skip. When they describe a problem, ask “Do you know what that is costing you?” That is how a vague complaint becomes a number in your proposal that they said, not one you estimated.

Prompt 2

Audit the transcript

This is the one that matters. Do not skip ahead to Prompt 3.

You are auditing a discovery call transcript. You are NOT writing a proposal yet.

HARD RULES
1. Never invent anything. If the transcript does not say it, you do not know it.
2. Quotes must be word for word. Never paraphrase inside quotation marks.
3. If you are unsure whether something was said, leave it out and list it as a gap.

TASK
Read the transcript below and produce this audit, in this order:

### Pain language
Exact quotes describing the problem. For each: [quote] - [who said it] - [which bucket: productivity, reputation, or profitability]

### Outcome language
Exact quotes describing what success looks like to them. Same format.

### Constraint signals
Budget, timing, policy, prior bad experiences, stated limits. Quote where possible.

### Trust signals
Who makes the decision. What they volunteered without being asked. How candid they were.

### Gaps
What a proposal needs that this call did not give me. List at least three. There are always at least three.

### The one line
If I could put a single sentence from this call into the first paragraph of the proposal, which is it, and why that one.

TRANSCRIPT
[PASTE HERE]

Prompt 3

Draft the proposal

Notice the input. You feed it the audit, not the transcript.

Write a first-draft proposal using the audit below.

HARD RULES
1. Use their words. If they said "I am the last stop on forty things a week," the proposal says that, not "internal review inefficiencies."
2. Never invent a fact. Anything the audit does not contain, mark inline as [GAP: what is missing].
3. Mark every inference. If the audit supports a conclusion but does not state it, write [INFERENCE: what you concluded and why] so I can accept or cut it.
4. Do not invent pricing. Use exactly what I give you below, or mark [GAP: pricing].

STRUCTURE
- What we heard. Their problem in their words, and their number.
- What success looks like. Their definition, not a generic one.
- What we will do. Scope, in plain language.
- What is included, and what prices separately.
- Timeline, anchored to the deadline they actually named.
- Investment.
- What happens next. Three concrete steps.

VOICE
Plain, direct, warm. Short sentences. No jargon. Nothing a person would not say out loud. Do not open by describing the market or the moment, open with them. Two pages, not forty.

MY PRICING AND SCOPE
[YOUR NUMBERS AND WHAT IS INCLUDED, OR WRITE "NOT SET" AND IT WILL MARK A GAP]

MY FORMAT (optional but worth it)
[PASTE A PROPOSAL YOU HAVE SENT BEFORE. Tell it: match this structure and section order, not the content.]

AUDIT
[PASTE PROMPT 2 OUTPUT HERE]
Do this one thing. Paste in a proposal you have already sent and tell it to match your structure. The draft stops looking like a generic template and starts looking like yours. It takes twenty seconds and it is the single biggest quality jump on this page.

Prompt 4

The human pass

Run this, then do the checklist yourself. The checklist is the real deliverable.

Review the proposal draft below as a skeptical editor. Do not rewrite it. Report findings only.

Check for and list:
1. Any claim about the client that does not trace back to the audit
2. Any remaining [GAP] or [INFERENCE] markers, listed so I can resolve each one
3. Sentences that sound like AI wrote them: throat-clearing openers, three-item lists in every paragraph, words nobody says out loud (leverage, robust, seamless, utilize, holistic), hedging like "may help to potentially"
4. Anywhere I described their problem in my words instead of theirs
5. Anything that reads as a promise or commitment I should double-check
6. Anything confidential about a third party who was not on the call

Then: the three specific edits that would most improve this, in priority order.

DRAFT
[PASTE HERE]

Before it goes anywhere

AI never decides your price, your scope, your timeline, or whether you actually want this client.

Sample discovery call. Yours to practice on.

This is a composite. It is not a real company. It is built from the shape of real engagements, with every name, number, and detail changed. Every signal from tonight is in it.

Read the transcript
The call: 34 minutes, video, recorded with permission. Second conversation, not the first. Seller: Dana Reyes, independent consultant. Buyers: Paula Merrow, COO and budget holder, and Nick Fitch-Alvarez, head of new business. Company: Merrow and Fitch, a 30-person regional insurance brokerage.

Paula: So that is the background. We did the workshop in the spring, people liked it, and then nothing. Which is the part that bothers me.
Dana: When you say nothing, what does nothing look like?
Paula: I mean I see two people using it. Maybe three. Nick uses it. Our marketing person uses it constantly. Everybody else went back to how they did it before. And I paid for licenses for the whole team, so that is money going out every month for software nobody opens.
Dana: How many licenses?
Paula: Thirty. Well, twenty-eight.
Nick: It is not that people do not want to. I think they do not know what to do with it on a Tuesday. The workshop was great but it was all here is what is possible. And then you are back at your desk with a renewal packet to get out and you just do it the way you know how.
Dana: That is the most common version of this I see. The gap is not interest, it is the Tuesday problem.
Paula: So what I keep going back and forth on is whether we need more training or whether we need something else. Because I do not want to buy another day of training and have the same thing happen. I would rather do nothing than do that again.
Dana: Say more about what something else might be.
Paula: I do not know. That is what I am asking you. Something that sticks.
Dana: Okay. If this works, what is different in six months? Not the tools. What is different for you.
Paula: I would stop being the bottleneck on everything that requires writing. Every proposal, every client letter, every Paula can you look at this before it goes out. I am the last stop on probably forty things a week and half of them do not need me.
Dana: Forty a week.
Paula: That is a guess but it is not a wild guess.
Nick: It is not a guess, it is low.
Dana: Nick, what about you? Six months out, what is different?
Nick: January. January is the whole thing for me. Open enrollment. We are at a normal pace now and then from the second week of November through the end of January it is insane. Last year I personally had about a hundred and ten renewal conversations in nine weeks. Every one needs a summary going in and a follow-up going out, and I do all of that myself at night.
Dana: At night.
Nick: At night. And the follow-ups are the part that slips. I know for a fact we lost two accounts last year that we would have kept if I had followed up inside a week instead of three.
Dana: Do you know what those two were worth?
Nick: Combined? Around forty thousand in annual commission. I would have to check but it is that neighborhood.
Dana: So the follow-up gap cost you roughly forty thousand dollars last year.
Nick: When you say it like that. Yeah.
Paula: That is. Okay. I had not heard that number.
Nick: I have said it.
Paula: You have said we lost two. You have not said forty thousand.
Dana: Nick, if I could give you one thing before November, one thing only, what is it?
Nick: The follow-ups. Not the summaries, I can do those fast. The follow-ups, where it pulls what we actually talked about and drafts the thing so I am editing instead of starting.
Dana: Paula, same question, one thing.
Paula: Getting the team to actually use it. Which I know is not a thing you can hand me in a box.
Dana: If we scoped something, who says yes?
Paula: Me. I would tell my partner but he is not going to block it. He does not have opinions about software.
Dana: And is there a budget you are working inside?
Paula: Yeah. I have got something in the range of, call it two to three thousand a month set aside for this kind of thing. Three is the top. I would want to feel really good about three.
Dana: That is helpful, thank you.
Paula: And I will be honest with you, the thing I am nervous about is not the money. It is, we are going to be paying you every month, and are we using it? Are people showing up? Because if I am writing that check and four people are coming to the sessions, that is on me to fix but I need to know it is happening.
Dana: That is a fair thing to be nervous about and I would rather build around it than pretend it will not happen.
Nick: When you say sessions. Is that everybody in a room, or is it?
Dana: It can be both. Typically a team-wide rhythm and then individual working sessions for the people with the most to gain.
Nick: Marisol.
Paula: Marisol, yeah. She would be the one.
Paula: What about the thing Nick wants, the follow-up thing. Is that part of it or is that separate?
Dana: That is usually a build, and builds price separately from a standing retainer. Otherwise the retainer quietly becomes a software project and the training stops happening.
Paula: That is fine. I would just want to know the number before it starts.
Dana: You would get a short scope and a price before any work begins. No surprise invoices.
Nick: Timing on something like that though. Because November.
Dana: Understood. Let me think about sequencing and put it in writing.
Paula: The other thing, and this might be a dumb question, is any of this a problem from a, we handle a lot of personal health information. Not medical records but enough. Is putting that into an AI tool a problem?
Dana: Not a dumb question, it is the right question, and the answer is it depends entirely on which tool and how it is configured. It is also something we would write down as a policy so your team is not guessing case by case.
Paula: We do not have anything written down.
Dana: Most firms your size do not. That is usually part of the first month.
Paula: Okay. Put that in there.
Dana: When would you want to start?
Paula: Soon. If we are doing it before Nick's busy season we are already a little late. October first feels right. Can you get me something I can look at this week?
Dana: Thursday?
Paula: Thursday works.
Nick: One more thing. Whatever you send, can it not be forty pages? I have gotten two of those.
Dana: It will not be forty pages.

A proposal is the smallest thing a transcript can do

I started recording calls so I could write better proposals. That is not what changed my business.

What changed it was running my own transcripts through AI and asking different questions. Where do I lose people. What do I over-explain. Which words do I reach for when I am not sure. What do I say when I am selling versus when I am teaching.

I learned more about how I communicate from six months of my own transcripts than from twenty years of being told I am a good communicator.

Other things to ask a transcript

  • What did I miss that they said twice?
  • Where did I talk more than they did?
  • What question did they dodge?
  • What would a skeptical buyer say is still unanswered?
  • What did I promise without noticing?

Same recording. Completely different questions. That is the part that compounds.

The tools I actually use

A couple of these are affiliate links, which means I get a small cut if you sign up. I would recommend them either way and I am telling you because you should always know.

Homework

Do this before you forget you know how

  1. Pick one discovery call you have already had. Get permission, or use the sample above.
  2. Run Prompt 2. Get an audit.
  3. Compare it against your own memory of the call. Find one thing you had forgotten. There will be one.
  4. Draft one section with Prompt 3. One section, not the whole thing.
  5. Run the checklist before any of it goes near a client.

Did something work?

If you use this on a real call, I would genuinely like to hear about it.

Can I share this publicly? Required

About Reed

Reed Jones is the founder and Chief AI Officer of The AI Department, where she helps women-led businesses and small teams put AI to work without losing their voice. She taught this class free for the Massachusetts AI Coalition. Broadcast background, which is why she cares how things sound.

Keep learning

Free, in person, same room

Paid workshops on Maven

See all workshops
Code BOSTON50 takes $50 off any of them. That one is for tonight's room.

Work with me

AI Fit Check, from $2,500.

Two weeks. Intake, a foundations session, interviews with your team, findings and a roadmap.

AI Build-Out, from $18,000.

Ten weeks. The full installation.

AI Advantage, from $4,500 a month.

Fractional AI leadership, three month minimum.

Work with The AI Department