The Call-Prep Workflow I Install First (Full Setup)
Every AI enablement engagement I run starts with the same install: call prep. Not because it's flashy — because every rep has a meeting tomorrow, and a prep doc they actually use within 24 hours beats any demo. This is the full setup, exactly as I install it. Copy it.
Why call prep first
Prospecting workflows look impressive in a slide. Call prep wins in the room.
A B2B SaaS team I worked with had already bought three AI tools and run two all-hands trainings. Usage still sat with the same three enthusiasts. We didn't start with outbound. We started with tomorrow's meetings. Within a week, managers were asking for prep docs in one-on-ones — not because we mandated it, but because the docs were short enough to read and specific enough to coach from.
That's the bar. If the first workflow doesn't change how a real call gets prepared this week, nothing downstream will stick.
The project instructions
Create a Claude Project called Call Prep — [your team] and paste this in, filling the brackets:
You prepare sales calls for [COMPANY], which sells [PRODUCT] to [ICP]. Our
sales process stages are [STAGES]. Our discovery methodology is
[e.g. "MEDDPICC" / "SPICED" / "our own"].
When given meeting context (account, attendees, stage, notes, emails),
produce a ONE-PAGE prep doc in EXACTLY this format:
## The one thing
The single most important thing to accomplish in this meeting, in one
sentence.
## Where we are
3–5 bullets: deal stage, what's been promised, what's changed since last
touch. Only from the context provided — mark anything assumed as "assumed".
## The room
Each attendee: role, what they care about, likely stance
(champion / neutral / blocker), and one thing to ask or confirm with them.
Unknown attendees: say "unknown — open with a role question".
## Questions that move the deal
Exactly 3 discovery questions, ordered. Each must target something we
genuinely don't know that changes the deal. No questions whose answers are
already in the context.
## Likely friction
Top 2 objections or awkward topics with a one-line honest response each.
## After the call
The follow-up the rep should be ready to send within an hour, in one bullet.
Rules: one page maximum. Terse bullets, not prose. Never invent history
that isn't in the provided context.
The daily prompt
Prep my [stage] call with [account] tomorrow. Attendees: [names/titles].
Here's the context: [paste CRM notes, recent emails, last call summary].
That's it. The fixed output format is doing the real work: every rep's prep doc looks the same, so managers can coach from them, and "show me your prep doc" replaces "did you prep?" in one-on-ones.
Annotated example prep
Here's what a good output looks like after the instructions are tuned. Context was a mid-funnel discovery call with a B2B SaaS account — CRM stage, two emails, and a short last-call summary pasted in. Comments in brackets are mine, not part of the doc.
## The one thing
Confirm whether Ops owns the evaluation budget this quarter, or whether
Finance has to approve before a pilot can start.
[Good: one decision that changes the deal path. Bad would be "build rapport"
or "cover the product."]
## Where we are
- Stage: Discovery / technical evaluation scheduled for next week
- Last call: champion (Director of Ops) agreed to bring Finance if budget
path was unclear
- Assumed: Finance attendee is new to the deal — not named in prior notes
- No pricing discussed yet; champion asked not to lead with pricing
## The room
- Jordan Lee, Director of Ops — champion; cares about onboarding time and
handoff between teams. Ask: what "done" looks like for a 30-day pilot.
- Sam Ortiz, Finance — unknown stance; likely cares about contract term and
who owns usage. Open with: role in this evaluation and approval path.
- Alex Kim, Solutions Eng (our side) — demo support; stay quiet until asked
for architecture detail.
## Questions that move the deal
1. Who has to say yes before a pilot can start, and whose calendar is the
bottleneck?
2. What would make you kill this evaluation in the next two weeks?
3. If the pilot works, what replaces the current process — and who loses
work they care about?
## Likely friction
- "We're comparing three vendors" — Honest: we expect that; ask what the
scorecard weights so we don't demo the wrong thing.
- "We need legal review before anything" — Honest: fine; ask what usually
slows legal so we can prep docs early, not react late.
## After the call
Send a 5-bullet recap with the decision owner named, pilot success criteria
in their words, and a proposed time for the technical session.
Notice what the instructions prevented: invented competitor names, fake history, essay-length context, and "great discovery questions" that the CRM already answered. The "assumed" flag is the trust feature — reps stop fighting the tool when it admits what it doesn't know.
Why the format is strict
Loose prompts produce essays nobody reads before a call. The one-page cap, the exactly-three questions, and the "assumed" flags exist because reps read prep docs in the two minutes before the meeting — in a parking lot, on a phone. Anything longer is shelf-ware.
Managers need the same shape every time. If every prep doc invents its own structure, coaching becomes "read this novel and find the insight." With a fixed format, a manager can scan five docs in five minutes and coach the "one thing" and the three questions.
What stays human
The conversation. The doc is a map, not a script — reps who read bullets off a page lose rooms. Stance labels are hypotheses to test, not facts to act on. Say this out loud when you roll it out; clear boundaries make people use the workflow more, not less.
Also human: deciding whether to send the follow-up as drafted, whether an objection response is honest enough for this account, and whether the "one thing" is actually the right bet. AI prepares. The rep owns the call.
First-session run-of-show (45 minutes)
I run the same session every time. No slide deck that outlives the hour.
0–5 min — Boundary, out loud. Nothing sends without a human review. Prep docs are for the rep, not the customer. Bad outputs are instruction bugs, not rep failures. Write that on a shared doc if you need to.
5–10 min — Paste the project instructions. Fill brackets live: product, ICP, stages, discovery method. Don't polish them in a back room first. The room should see the system get born.
10–35 min — Every rep preps a real call this week. Never demo data. Each person pastes CRM notes and emails for a meeting they actually have. Run the daily prompt. Read the output together for two or three people. Fix the instructions when something is wrong — too long, too generic, inventing context, weak "one thing."
35–45 min — Manager loop. Agree that next week's one-on-ones include "show me your prep doc." Name one person who will collect the first week's best examples and drop them in the team channel. End with tomorrow's meetings already prepped.
Watching the system get corrected in the room is what convinces skeptics it's steerable rather than magic.
Failure modes I see in the first two weeks
Context dumps without a stage. Reps paste a whole email thread and get mush. Fix: require stage + attendees + "what's changed since last touch" in the daily prompt before the paste.
The "one thing" is soft. "Learn more about their process" is not a one thing. Fix: add a line to the instructions — the one thing must be a decision, confirmation, or commitment that changes the deal path.
Questions that restate known facts. If the CRM already says budget is Q3, don't ask about budget timing. Fix: reinforce "no questions whose answers are already in the context" and delete those questions live in the first session so people see the rule enforced.
Managers never ask. Usage dies by Friday. Fix: put prep docs on the one-on-one agenda before the install session ends. Adoption follows inspection.
Docs get longer every week. Someone "improves" the instructions with extra sections. Kill them. One page is the product.
Day-30 check
At day 30 I ask four questions. Anything that fails gets fixed or killed.
- Would reps complain if we removed this? If no, it never became a habit — reinstall with real calls, or stop pretending.
- Are managers citing prep docs in coaching? If no, the inspection loop never started.
- Are outputs still one page? If no, instructions drifted — restore the format and delete the extras.
- Has anyone shipped a second workflow on top of this? Call prep is the trust layer. If it stuck, research and follow-up drafts are next. If it didn't, don't add more.
A healthcare wellness org I worked with passed all four by week five. The signal wasn't license count — it was managers saying "your prep doc said the one thing was budget owner, and you spent the call on product tour." That's adoption: the workflow changed how work got coached.
Soft CTA
This is one of ten installs in the GTM AI Operating System — the full revenue cycle from prospect research through forecast narrative, plus the operator layer that keeps adoption alive past week two: rollout guide, boundary doc, adoption tracker, and the model-update runbook. The OS is this post, times ten, with the system around it.
If you want it installed with your team rather than handed as docs, the GTM AI Sprint is the live version of this session — real accounts, fixed instructions, manager loop included.
Frequently asked
Highest immediate value, lowest risk. Every rep has a meeting tomorrow, the output is for the rep's eyes only, and a good prep doc within 24 hours is the fastest way to build the trust that every later workflow depends on.
No. This install runs in a plain Claude Project with pasted context — CRM notes, emails, attendee lists. Connectors make it faster later, but the workflow works on day one without any integration.
You can see signal in the first session. The real check is day 30: would reps complain if you removed it? If managers aren't asking for prep docs in one-on-ones by then, the install didn't land.