Practice the call before the call.
Turns a deal, framework, or objection into a live AI roleplay partner, then scores the rep against the framework with a cert bar.
Inside: The buyer persona, the real objections, and a framework-scored, pass-or-not-yet scorecard.
Install it in one line, or paste it in.
~/.claude/skills/ and runs automatically when it is relevant.Connect your context. Set it to your motion.
it builds the persona from a real open deal, the stakeholders, the stage, the stated problem, and the notes and activity your reps actually logged, so the buyer is this account, not a generic one.
it pulls the objections your buyers actually raise, so reps rehearse the real ones, not generic ones.
the real back-and-forth with the account, so the buyer in the roleplay sounds like the one in the thread, tone and all.
how the account actually uses you, so the value story and the objections are grounded in real usage instead of guesses.
it saves the scenario and the scorecard as a reusable module, so a good roleplay becomes a certification other reps can take.
This was built for a B2B SaaS team running a named methodology. Set these to your motion:
| Set this | What it is | Default / Example |
|---|---|---|
| MOTION | the conversation being practiced | discoverydemoobjectionrenewalreach-up |
| FRAMEWORK | what the rep is graded against | SPRINT discovery →CLEAN demo →LEARN objections →PLAN →VALUE → |
| PERSONA | who the buyer is | the seattheir concerntheir pressure |
| DIFFICULTY | how hard the buyer pushes | warmneutralskepticalhostile |
| CERT_BAR | the score a rep must clear to go live | a passing scorecard on the framework |
Everything the skill does, in full.
Takes a conversation a rep is about to have, discovery, demo, an objection, a renewal, a reach-up to the economic buyer, and turns it into a live roleplay they can practice against. It builds the buyer persona with a real concern, runs the scenario in character, throws the objections that seat actually throws, and then grades the rep against the framework with a specific, coachable score. It turns "read the deck and wing it" into reps you can measure, so the first live attempt is not the first attempt.
- 1Build the buyer, not a cardboard cutout
A useful roleplay starts with a real persona: the seat, the concern that seat actually carries, and the pressure they are under. A CFO stalls on a different thing than a champion does. The buyer has a point of view and defends it, so the rep has to earn the room.
- 2Run the scenario in character
The skill plays the buyer and stays in character through the call. It does not hand the rep the answer. It pushes where a real buyer pushes, goes quiet where a real buyer goes quiet, and raises the objection at the moment it actually lands.
- 3Throw the objections they will really hit
Objections are drawn from the motion and, when connected, from your real calls: price, low usage, "we already have this", a champion who left, no compelling event. The rep practices the turn (LEARN: listen, empathize, anchor, respond, next step) against the ones that actually show up.
- 4Score against the framework
When the roleplay ends, the rep gets a scorecard tied to the framework, not a vibe. Each element scored, the specific moment it was won or missed, and the one thing to fix before the live call. A cert bar makes it a pass or a not-yet, so "ready" is a decision, not a hope.
- 5Save it as a reusable module
A good scenario is worth running more than once. Saved, it becomes a certification other reps take, so enablement is a library that compounds, not a one-off session that evaporates.
- The buyer stays in character and never coaches the rep mid-scene. The lesson is in the score, not the scene.
- Objections are real to the motion, and grounded in actual calls when a call tool is connected, never generic filler.
- The score is tied to the framework element by element, with the specific moment cited, not a single number.
- The cert bar is an explicit pass or not-yet, so readiness is decided, not assumed.
- No buyer detail is invented past what you gave or what a connected record shows. A gap is named.
Where an operator takes this next.
This runs today as a text roleplay. Here is where an operator takes it next, and none of it is science fiction.
The skill is the seed. The build is yours.
Wire the scenario to an ElevenLabs voice agent so the rep practices out loud, against a buyer that interrupts, stalls, and pushes back in real time. The scorecard still grades the framework; the rep just rehearses with their voice, not their keyboard.
A small tool, vibecoded in a day in Lovable and shipped on Vercel, pulls every rep's recorded calls from Gong, finds the objections and the moments they lost, and generates next month's roleplay slate automatically, each scenario tuned to the exact gap that rep keeps hitting.
Turn that into a monthly plan per rep: the tool reads the calls, writes the plan, and hands each rep an ElevenLabs agent loaded with their accounts and their weak spots, so practice is personal and on the calendar, not a generic SKO.
One skill is the on-ramp.
A single skill does one job. Chained into a playbook, or run as a full build, it becomes a system. Here is where this one plugs in.