Build a 5-minute pre-call brief: goal, hypotheses, questions, and unknowns

What this does

You paste what you know. The AI returns a one-page call plan: goal, agenda you can propose, hypotheses to test, questions in order, likely objections, and the gaps in your information.

Who it’s for

AEs and SDRs with a call or demo coming up and little time. CSMs preparing for a business review. Not a replacement for reading the account notes yourself.

What you need

• Meeting type and length
• Attendees and their roles (initials or titles if you want privacy)
• What you already know (notes, prior emails summarized)
• Your goal for the call and the outcome you’d call a win
• Tool: Any AI chat tool (for example ChatGPT, Claude, Gemini, or Grok)
• Cost: $0 with a free AI chat tier (limits change)
• Setup: 2 minutes (estimate)

Steps

1. Gather what you know in rough notes: who’s attending, what they said before, what they use today.
2. Fill the [VARIABLES] and run the prompt.
3. Delete any hypothesis that has no support in your notes, or mark it ‘to test.’
4. Copy the questions into your call notes in the order you’d ask them.
5. After the call, add what you learned to the unknowns list; those become your follow-ups.

The prompt

You are helping a salesperson prepare for a call. Use ONLY the notes I provide. If something isn't in my notes, label it "Unknown." Do not guess facts about the company, the attendees, or their tools.

NO-ADDITIONS RULE: Do not add facts, tips, statistics, scarcity or urgency, promises, references or testimonials, or the gender of any person that I did not supply.

Meeting type and length: [e.g. 30-min discovery, 45-min demo, 20-min renewal check-in]
Attendees and roles: [ATTENDEES]
What I know so far: [NOTES]
What I sell: [PRODUCT, ONE SENTENCE]
My goal for this call: [GOAL]
A good outcome would be: [NEXT STEP]

Write a one-page call plan:
1. Call goal (1 sentence) and a proposed agenda I can say aloud in 20 seconds
2. What I know (bullets, from my notes only)
3. Hypotheses to test (label each "Hypothesis"; say what answer would confirm or disprove it)
4. Questions in order (8 or fewer, open-ended, grouped: situation, problem, impact, next step)
5. Likely objections and a short honest reply to each. Don't invent claims about my product; if you need a fact about it, write [ADD FACT].
6. Unknown list: what I should find out on the call
7. Next-step ask I could make at the end

Plain words. Under 400 words.

Sample input (fake data)

All names and companies are fictional. This is the input for run 4 (the output below), on the current prompt. Runs 5 and 6 used different inputs; see the test log.

Meeting: 30-min discovery
Attendees: Director of Purchasing and a Warehouse Supervisor (fictional company: Harlow Building Supply)
What I know: Purchasing downloaded our guide on reducing reorder errors. They have two yards. No other info.
What I sell: Inventory reorder-alert software
Goal: Understand how reorders are triggered today
Good outcome: A second meeting with the person who approves software

Sample output

Real output from run 4 on 2026-10-14, using the current prompt, from an unspecified AI assistant (Grok Bot agent session; model name not shown). Copied as produced. Nothing was fixed after the fact; the problems we found in it are listed under “How it went”.

**1. Goal and agenda**
Goal: Learn how Harlow decides when to reorder and who approves a change.
Say aloud: "I'd like to spend a few minutes on how reorders get triggered across your two yards, what goes wrong, and who else would be involved in a change. Then we can decide on a next step. Does that work?"

**2. What I know**
- Purchasing downloaded our guide on reducing reorder errors.
- They have two yards.
- Unknown: everything else.

**3. Hypotheses to test**
- Hypothesis: Reorder errors are a real problem for them. Confirm: they describe a recent error. Disprove: they only downloaded the guide out of curiosity.
- Hypothesis: Each yard reorders separately. Confirm: two people or two lists. Disprove: one central list.
- Hypothesis: The Director of Purchasing approves software. Confirm: they say so. Disprove: they name someone else.

**4. Questions**
Situation: How does a reorder get triggered at each yard today?
Situation: Who places the order?
Problem: What did you want to fix when you downloaded the guide?
Problem: What happens when an item runs out?
Impact: How often does that happen, in your experience?
Next step: Who approves a new tool like this?
Next step: Could we set 30 minutes with them?

**5. Likely objections**
- "We handle it with spreadsheets." Understood. What would need to change for that to stop working?
- "What does it cost?" [ADD FACT: pricing]
- "How long does setup take?" [ADD FACT: setup time]

**6. Unknown**
Current process, error history, who approves, budget, timing, other tools.

**7. Next-step ask**
"Could we set 30 minutes with whoever approves software, so I can answer their questions directly?"

Check before you send

• Section 2 only lists things from your notes.
• Product facts in objection replies are true; fix any [ADD FACT].
• Questions fit the time you have.
• No hypotheses sound like facts.

How it went

• Test date, tool, model: 2026-10-14. Unspecified AI assistant (Grok Bot agent session; the model name is not shown to the agent). One tool only.
• Runs logged: 6, on 6 different fake inputs (see the test log). Runs 1-3 used the original prompt. Runs 4-6 are a second round on new fake inputs after the prompt was fixed, as listed below. Every run was done by the same AI assistant session that wrote the fake input, and in round 2 the same session also edited the prompt, so these are not blind or independent runs. A script counted words and characters against the prompt’s limits. No second person re-ran it from the written steps.
• Why the label is “Tried in an AI tool,” not “tested”: we reserve “tested” for plays with at least 3 logged runs in at least 2 different AI tools, and we ran one tool. We also have no independent reproduction of the steps and no timed manual baseline. The runs are real, but this is one tool’s behavior, and 3 runs per round is too few to say a problem is gone.
• Time saved: Not measured. We did not time a manual baseline or an assisted run.
• What we changed after round 1: Added: no-additions rule only.
• Round 2 (current prompt, runs 4-6). What worked:
  • All 3 plans were 264-278 words (limit 400), with all 7 sections and 7 questions. A script found no numbers or gendered pronouns that weren’t in the notes.
  • Section 2 stayed notes-only. Product facts were [ADD FACT], not invented.
  • In run 5 the model marked a hypothesis with ‘No current usage data in my notes’, which is the behavior the no-additions rule asks for.
• Round 2: what failed or needed fixing:
  • Run 6 (thin notes) is still generic. The rule stops invented facts but doesn’t stop generic hypotheses.
  • Questions come grouped by label, not as a numbered list, so counting ‘eight or fewer’ took a script (same as the first round).
  • Run 4 hypothesis 1 (‘reorder errors are a real problem for them’) leans on a guide download; it is labeled a hypothesis with a way to disprove it.
• Checked against “Check before you send”: Section 2 notes-only in all 3 runs; [ADD FACT] gaps remain to fill; question count fit; no hypothesis was written as a fact.
• Edge cases: Thin notes (run 6) unchanged. No prompt change was aimed at it.
• Round 1 (original prompt, runs 1-3): what failed, kept for reference:
  • Run 3 (thin notes): the hypotheses (pricing question means comparing options, Controller owns process, manual receipts are the pain) are plausible but rest on one form fill. The questions would fit any software.
  • Run 2 hypothesis 3 (‘Reminder texts still deliver value even if logins fall’) is explicitly a guess with no data behind it. A rep reading fast could say it to the customer as fact.
  • The questions came grouped as paragraph lines, not a numbered list, so counting ‘eight or fewer’ took a script.
• Not yet observed: how a different model or tool handles the same prompt (a second-tool run is still to do).

Watch-outs

Privacy: don’t paste confidential customer information into a consumer AI tool unless your policy allows it; use initials or titles. Don’t feed it recorded-call transcripts unless you had consent to record (see Safe Use).

See the Safe Use page: AI drafts, you decide, you send.

Variations

• Renewal version: add ‘usage notes and issues’ to the input and ask for risks to the renewal.
• Manager version: ask for ‘three things to coach the rep on before this call.’
• Demo version: add the features you plan to show and ask for the order that fits their pains.

Sources

None. Original wording. (Add any public framework or post used, with link and checked-on date.)

Retest

Re-check by: 2026-12-13 (60 days after the 2026-10-14 runs), or sooner when run in a second tool.

Test log

Changelog

• 2026-10-14: Draft written. Not run yet.
• 2026-10-14: Ran 3 times in one AI assistant on fake data. Filled in sample input and output, how it went, and the test log. Front matter set to runs_logged 3, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested. The prompt text was not changed after the runs.
• 2026-10-14: Prompt revised after round 1. Added: no-additions rule only. Re-ran 3 fresh runs (4-6) on new fake inputs in the same AI assistant. Updated the sample input and output to run 4, the “How it went” section, and the test log (original runs 1-3 kept, marked). Front matter set to runs_logged 6, last_tested 2026-10-14, retest_by 2026-12-13. The 2-tool bar for the word “tested” is not met, so the play is not labeled tested.
• 2026-10-15: Published with the evidence label “Tried in an AI tool” (6 runs, 1 tool, fictional inputs). Earlier wording that implied more evidence than one tool’s runs was replaced with the label; the runs, results, and limits above are unchanged.