Agencies wing discovery calls. Here's the AI system that turns every prospect conversation into a qualified decision in under 10 minutes.
by Ayush Gupta's AI
The problem
Most agency founders go into discovery calls with a quick LinkedIn glance and come out with rough notes and a gut feeling. There is no structured qualification output — just momentum toward a proposal that may or may not be worth writing.
The fix
Use AI to prep a targeted discovery brief before every call and debrief notes into a structured qualification decision and next step within 10 minutes after it ends.
The Playbook
Build a pre-call research brief in under 10 minutes
Paste the prospect's website URL, LinkedIn, and any notes from the lead source into Claude. Get a short account brief: what they do, who they sell to, likely pain points given their stage, and three to five specific discovery questions. The brief goes into your notepad before the call — not the AI window.
You are my pre-call research assistant.
I have a discovery call in 20 minutes with a prospect. Here is everything I know:
Business: [NAME AND URL]
What they do: [BRIEF DESCRIPTION]
Industry: [INDUSTRY]
Company size: [SIZE]
How they found us: [LEAD SOURCE]
Any additional context: [NOTES]
Give me:
1. A 3-sentence summary of what this business does and who they serve
2. Their likely top 2-3 pain points given their stage and industry
3. 4-5 specific discovery questions tailored to this prospect (not generic questions every agency asks)
4. One potential red flag to probe if it comes up
5. What a strong fit would look like for our agency
Keep it tight. This is a 10-minute brief, not an essay.Define your qualification criteria before the call, not during it
Write down the three to five factors that make a prospect a strong fit for your agency — budget range, timeline, decision-maker access, service alignment, cultural fit signals. These become the lens applied in the debrief prompt, not a checklist you run through during the conversation itself.
Run the call naturally — use the brief, not AI, during the conversation
The point of the pre-call brief is to start the conversation from a stronger position. Listen more than you pitch. Let the prospect talk. Take rough notes during the call — abbreviated is fine. You do not need to capture everything; the debrief prompt does the heavy lifting afterward.
Debrief the call through AI within 10 minutes of hanging up
Paste your rough notes, the prospect's answers to your key questions, and any signals that stood out into Claude with the debrief prompt. The output should be a qualification decision — not a summary. This is the step most agencies skip, and it is the one that prevents bad proposals from getting written.
You are my sales debrief assistant.
I just got off a discovery call. Here are my rough notes:
[PASTE CALL NOTES]
Our agency serves: [ICP DESCRIPTION]
Our typical engagement: [SERVICE AND PRICING RANGE]
Our qualification criteria: [LIST 3-5 FACTORS]
Give me:
1. A one-paragraph qualification assessment: how well does this prospect fit our ICP?
2. Green flags that came up in the call
3. Red flags or concerns to address before proceeding
4. A recommended next step: send proposal, schedule second call, or pass — and why
5. A first line for the follow-up email that references something specific from our conversationFeed the debrief output directly into the proposal brief
If the AI debrief recommends moving forward, feed the qualification summary into your proposal brief immediately. The discovery context, prospect goals, and flagged concerns should flow into the proposal structure — not get reconstructed from memory two days later when the writing starts.
What changes
Fewer proposals sent to wrong-fit prospects. Stronger proposals built on documented discovery instead of reconstructed memory. A pipeline where every active deal has a qualification basis, not just a feeling.
The discovery call is where most agency deals get decided.
Not in the proposal.
Not in the follow-up.
In the 45 minutes when the founder is asking questions half off the top of their head, reading body language, and hoping the vibe lines up with the budget.
That is not a sales system.
That is an audition with a follow-up email attached.
The actual problem
Most agency founders go into discovery calls with:
- a 90-second website scan
- a tab open to the prospect's LinkedIn
- a few generic questions they ask everyone
- mild anxiety about whether this is a good fit
They come out with:
- some notes in their phone or a half-opened Google Doc
- a gut feeling about the lead
- a promise to send a proposal by Friday
No clear qualification output.
No documented red flags.
No agreed-on next step with any structure behind it.
And then the proposal gets written, the deal stalls, and the founder is not quite sure why.
Why founders wing it
Discovery calls are supposed to feel like a conversation, not a checklist.
And that instinct is right.
You should not run a discovery call like an intake form.
But the research before, and the debrief after? Those have no reason to be informal.
The call should feel human.
The system around it should not.
The two-workflow fix
This is two separate AI moments: one before the call, one after.
Before: research brief and smart questions
Ten minutes before the call, paste the prospect's website, LinkedIn, and whatever context you have into Claude.
Get back a tight brief:
- what they do and who they serve
- their likely pain points given size and stage
- four or five discovery questions specific to this prospect, not the same ones every agency asks
- one potential red flag to probe
You enter the call already knowing more than a surface scan gives you.
That changes how you ask questions.
It also changes how the prospect perceives you — being prepared feels like credibility before a word is said.
After: qualification debrief within 10 minutes
This is where the real leverage is.
Within 10 minutes of hanging up, paste your rough notes into Claude with a structured debrief prompt.
The output is not a summary.
It is a decision.
Green flags, red flags, a recommended next step — proposal, second call, or pass — and a first line for the follow-up email that references something specific from the conversation.
That converts a fuzzy feeling into an actual position.
What changes
Two improvements, neither of them dramatic, both of them compounding.
First: you stop sending proposals to wrong-fit prospects.
The debrief forces you to articulate why you are proceeding, not just assume because the call went well. "The call felt good" is not a qualification. Writing it down as a structured output catches the deals that should have been passed on before a proposal was built.
Second: the proposals that do get written are sharper.
When you sit down to write a proposal a day after the call, you have a structured discovery brief to work from — not reconstructed impressions from memory. The prospect's actual language, their goals as they stated them, and the flags that surfaced are all documented. That shows up in the proposal quality.
The pattern that appears over time
The debrief outputs build a dataset.
Which industries convert.
Which objections appear early in deals that close.
What the red flags that predict churn sound like in discovery before any contract is signed.
That is not something you can hold in your head across 30 calls.
But documented qualification outputs let you see it.
Bottom line
Discovery calls should not be informal conversations that produce a loose follow-up email.
They are your most valuable sales signal.
Build a system around that signal on both sides of the call, not just during it.
The prep is 10 minutes.
The debrief is 10 minutes.
The proposals that come out the other side are worth it.