How to Turn Discovery Calls into Proposals with AI

The Problem: Why Discovery-to-Proposal Takes So Long

The typical agency discovery-to-proposal workflow has five manual bottlenecks. First, you're trying to take notes while also running the conversation — which means you miss details. Second, after the call you spend 20–30 minutes reconstructing what was said. Third, you manually extract scope, budget, timeline, and deliverables from scattered notes. Fourth, you open a blank document and draft the proposal from scratch. Fifth, you format, review, and send — usually the next day.

Each of those steps is a time sink, but the real cost is the delay. A proposal sent 48 hours after a discovery call converts at a measurably lower rate than one sent same-day. Speed signals competence. It tells the prospect you have systems, not chaos.

The workflow below replaces most of the manual work with AI-assisted steps. You still review everything — AI doesn't replace your judgment — but it eliminates the blank-page problem and the reconstruction work that eats most of the time.

The Workflow at a Glance

The full flow has five steps. Each one feeds into the next, and the entire sequence can run in under 45 minutes once you've done it twice.

1. Record and transcribe the discovery call automatically
2. Extract key information from the transcript using a structured AI prompt
3. Generate a proposal draft from the extracted data using a second prompt
4. Review and personalize — fix what AI gets wrong, add what only you know
5. Send and follow up — same day, while context is fresh

The time breakdown: Step 1 is passive (the tool records while you talk). Step 2 takes about 5 minutes. Step 3 takes 5 minutes. Step 4 — your review — takes 15–25 minutes. Step 5 takes 5 minutes. Total active time: roughly 30–40 minutes.

Step 1: Record and Transcribe the Call

Stop taking notes during discovery calls. Your job on the call is to listen, ask good questions, and build rapport. Let a recording tool handle the capture. Every major video conferencing platform now supports AI meeting assistants that join the call, record audio, and produce a transcript within minutes of the call ending.

Set your tool to auto-join all scheduled calls, or manually invite it when a call starts. After the call, you'll have a full transcript — usually within 2–5 minutes. Most tools also generate an automatic summary, but don't rely on that for proposal work. The generic summary misses the specific details you need. You want the full transcript for the next step.

One important note: always tell the prospect you're recording. A simple “I'm going to have our note-taking tool join so I can focus on the conversation instead of typing” works well. Nobody objects to this — it actually builds confidence that you're organized.

For a detailed comparison of these tools, see the best AI meeting assistants for agency teams.

Step 2: Extract Key Information with AI

Once you have the transcript, the next step is pulling out the specific details you need for a proposal. Don't read the whole transcript manually — paste it into Claude or ChatGPT with a structured extraction prompt. The prompt tells the AI exactly what to look for and how to format the output.

This step takes about 5 minutes: copy the transcript, paste it with the prompt, review the extracted output, and correct anything that looks off. The AI will occasionally misattribute who said what or infer scope details that weren't explicitly stated. Skim the output against your memory of the call before moving on.

The output gives you a structured brief that would normally take 20–30 minutes to compile from notes. Save this output — it feeds directly into Step 3, and it's also useful as a CRM note or internal handoff document if someone else on your team will work on the project.

Step 3: Generate the Proposal Draft

With the extracted information in hand, you now generate a first draft of the proposal. The key here is giving the AI a clear structure to follow — your agency's proposal format — along with the extracted data from Step 2. The AI fills in the sections; you refine them in Step 4.

Don't ask the AI to “write a proposal.” That produces generic, bloated output. Instead, give it the exact sections your proposal should contain and let the extracted data drive the content. The prompt below uses a standard agency proposal structure, but you should modify the sections to match whatever format you already send to clients.

This draft will be 70–80% ready. The structure will be right, the scope section will reflect what was actually discussed, and the language will be professional. What it won't get right: pricing, nuanced positioning, and anything that requires reading between the lines of the conversation. That's what Step 4 is for.

Step 4: Review and Personalize

This is where your expertise matters most, and it's the step you should never skip. The AI draft gives you structure and a first pass at content. Your job is to make it accurate, specific, and human. Plan to spend 15–25 minutes here.

There are five things AI consistently gets wrong in proposals that you need to fix every time:

• Pricing — always fill this in yourself. AI has no context for your rates, margins, or how you value the work.
• Scope boundaries — AI tends to be vague about what's included vs. excluded. Tighten this. Ambiguous scope is where projects go sideways.
• Client-specific nuance — if the prospect mentioned a past bad experience with another agency, or a specific internal constraint, weave that in. The AI might have captured it in the extraction but won't know how to use it strategically.
• Tone calibration — some clients want formal, some want casual. Adjust to match how they spoke on the call.
• Over-promising — AI drafts tend to be slightly more optimistic than you should be. Trim any deliverables or timelines that feel aggressive.

After your review, paste the proposal into your standard template — whether that's a Google Doc, Notion page, or dedicated proposal tool like PandaDoc or Qwilr. The AI draft is raw content; your template adds branding, formatting, and a professional finish.

Step 5: Send and Follow Up

Send the proposal the same day as the call. This is the whole point of the workflow — compressing a 2-day process into a same-day delivery. A proposal that arrives within hours of the conversation makes a strong impression and keeps the momentum alive.

Write a short cover email that references something specific from the call — not a generic “thanks for your time.” Mention a detail that shows you listened. Then attach or link to the proposal and state a clear next step: “Let me know if you'd like to discuss any of the scope details. I'm available Thursday or Friday for a quick follow-up.”

Set a follow-up reminder for 3 business days out. If you haven't heard back, send a short check-in — not a “just following up” email, but something that adds value: a relevant case study, a clarification on something discussed, or an answer to a question they raised on the call.

The Full Prompt Pack

Below are all the prompts used in this workflow, ready to copy and paste. Save these in a shared doc your team can access. Modify the proposal structure to match your actual format — the prompts work best when they reflect how your agency already writes proposals.

Related resources

Making This Your Default Workflow

The first time you run this workflow, it will take about an hour as you get familiar with the prompts and adjust them to your proposal format. By the third time, you'll be under 45 minutes. By the fifth, it'll feel automatic.

The key to making it stick: save your customized prompts somewhere your whole team can access them. A shared Notion page, a Google Doc, a Slack canvas — it doesn't matter where, as long as everyone on the team uses the same prompts. Consistency is what turns this from a hack into a system.

Start with your next discovery call. Record it, run the extraction prompt, generate the draft, review it, and send it same-day. One complete cycle will show you exactly where to customize the prompts for your agency's specific needs.

This workflow connects to a broader system for automating client onboarding. Once the proposal is signed, the client onboarding AI checklist picks up where this guide leaves off.