Discovery-to-Proposal Prompt Chain: One Call Transcript, Three Prompts, a Finished Cover Letter

What This Builds

This is not the reusable Custom GPT that stays loaded with your SOW template and rate card for every future deal. That build runs once per engagement, fed by an already-completed intake form, and produces a formatted SOW. This chain is different on purpose: it's for the raw transcript of one discovery call, before an intake form even exists, and it runs in a single sitting inside one ChatGPT conversation. It also goes further than opening the transcript to get a single scope draft in one pass. Here you extract ranked pain points first, generate two scope options next, and only then draft a cover letter, so you choose between options before committing to language a prospect will read.

Prerequisites

• Plus subscription ($20/month) or the free tier, since this chain runs as three prompts in one ordinary conversation
• A discovery call transcript or your own detailed notes from the call
• Time set aside to redact the transcript before pasting it in, since this step can't be skipped

There's no new subscription cost here beyond whatever ChatGPT plan you already use day to day.

The Concept

Think of this chain as three short interviews with the same document. First you ask what the prospect's problems are. Then you ask what it would take to fix them. Finally you ask for a short note pointing at the option you picked. Each answer feeds the next question inside the same conversation, so ChatGPT never loses the thread between steps the way it would if you started three separate chats.

Build It Step by Step

Part 1: Redact the transcript

Before pasting anything, strip financial figures the client mentioned, any credentials or account details, and third-party names not central to the engagement. This step protects the client and protects you; a discovery call transcript should never carry more detail into a chat window than the chain actually needs to rank pain points and draft a scope.

Part 2: Run the three prompts in order

Prompt 1, pain points:

Here is the transcript from a discovery call with a prospective client. Read
it and list their top 5 to 7 pain points, ranked by how much each one costs
the business in time, risk, or money. For each pain point, quote the
specific line from the transcript that supports it.

Prompt 2, scope options:

Using the ranked pain points above, draft two scope options. Option A
addresses only the top 2 to 3 pain points. Option B addresses all of them.
For each option, list the deliverables and a rough timeline in weeks. Do
not include pricing.

Prompt 3, cover letter:

Using Option [A or B], draft a two-paragraph proposal cover letter to the
client. Reference their specific pain points by name from the list above,
and set up the attached scope of work without repeating it line by line.

Part 3: Test and refine

Check that every pain point in the Prompt 1 output carries an actual quoted line. A pain point without a quote is a sign ChatGPT inferred something the prospect didn't say, and that's worth catching before it shapes a scope option. If Prompt 3's tone reads too promotional, add a line to the prompt telling it to write like a colleague explaining a plan, not a salesperson closing a deal.

Real Example: A Law Firm's First Discovery Call

Setup: A redacted transcript from a discovery call with a 15-person law firm.

Input, Prompt 1: The full redacted transcript.

Output, Prompt 1: Six ranked pain points, including a slow, aging file server and no multi-factor authentication on email accounts, each with a quoted line from the call.

Input, Prompt 2: No new input needed. Just the second prompt in the same conversation.

Output, Prompt 2: Option A covering the file server and MFA rollout over four weeks. Option B adding a backup overhaul and a Microsoft 365 licensing review over eight weeks.

Input, Prompt 3: "Using Option B" plus the third prompt.

Output: A two-paragraph cover letter naming the file server slowdown and the MFA gap directly, then pointing to the attached eight-week scope.

Value: Keeps one engagement moving through three drafting steps in a single sitting instead of returning to a blank page for each document separately.

Before It Goes to the Prospect

Read all three outputs end to end against the original transcript before anything leaves your inbox. Confirm the scope actually matches what you'd deliver, fix any pain point that doesn't hold up under a second read, and only then paste the cover letter into your own email to send. This chain drafts three documents. It doesn't send any of them, and it shouldn't; that decision stays with you.

What to Do When It Breaks

• ChatGPT invents a pain point the prospect never actually raised → Require a quoted line for every pain point in Prompt 1's output, and drop anything that arrives without one.
• The conversation loses track of which scope option you picked → If the chat has gotten long, start fresh and paste a short summary of the chosen option before running Prompt 3.
• The cover letter reads generic despite specific inputs → Add an instruction telling it to name the client's stated pain points explicitly rather than summarizing them abstractly.
• You realize halfway through you should have picked the other option → Re-run Prompt 3 with the other option's letter, referencing "Option A" instead. The earlier steps stay in the conversation history for reference.

Variations

• Simpler version: Skip the two-option split in Prompt 2 and generate a single scope every time, if your engagements tend to be similar in size.
• Extended version: Once you have the ranked pain points and chosen scope from this chain, feed them as the intake form input to a Custom GPT already loaded with your firm's SOW template and rate card, turning this chain's output into that build's starting point.

What to Do Next

• This week: Run the chain live right after your next discovery call, while the details are still fresh.
• This month: Build a short redaction checklist so Part 1 takes minutes instead of guesswork each time.
• Advanced: Track which pain points show up most often across prospects, and use that pattern to sharpen the discovery questions you ask on the next call.

Advanced guide for IT consultant professionals. These techniques use more sophisticated AI features that may require paid subscriptions.