Welcome back, Agent 👋
Pick a way to train. Everything here is safe practice — no real leads are called.
Objection Trainer
Face one objection at a time. Respond, then get instant scored coaching.
Full Role-Play
Run a full appointment-setting call or presentation with a live AI client.
Live Client Mode
Enter a real client's basic info and rehearse that specific appointment.
📖 How to use this trainer (quick start)
- Pick a mode above. Objection Trainer = drill one objection. Full Role-Play = a complete simulated call. Live Client Mode = rehearse a specific real appointment you have coming up.
- Choose a scenario and difficulty, then click Start. You'll hear a phone ring and the AI client will speak first — just like a real call.
- Reply two ways: type your line in the box and click Send, OR talk — tap 🎤 Speak, say your whole line, then tap it again (it now says "tap to send") to send it. The client answers out loud. First time you use the mic, your browser asks "Use your microphone?" — click Allow. On a downloaded file it may ask each turn; that's normal.
- Get coached any time. Hit Coach my last answer after any line — you don't have to finish the call. You'll get a 0–100 score, what worked, what to sharpen, and a better way to say it.
- ⏸ Pause freezes the call; ▶ Resume continues. End & get scorecard gives a full review of the whole call.
The golden rules the coach is looking for: open low and slow like a doctor's-office receptionist (calm, just verifying info — never excited or salesy), and dig into the emotional "why" — what happens to their family if they don't have this. No "why," no sale.
📚 Study & Train
Brush up on a skill first, then jump straight into a role-play to practice it. Pick a topic.
🎯 Objection Trainer
Choose an objection. The client will hit you with it — respond in your own words, then get scored coaching.
💬 A real client hit you with something else?
Type exactly what the client said. The coach will give you the recommended response from the playbook — and you can practice that exact objection live before you call them back.
🎭 Full Role-Play
A complete simulated call. Pick the call type — the client's situation is a surprise that changes every session, so you learn to think on your feet.
🎲 A hidden call situation (new lead, aged lead, hostile answer, equity/critical-period, unhealthy → beneficiary pivot, already-covered reset, and more) is chosen at random each time you start — so you won't know which is coming. Every one still hinges on uncovering the why.
Type your line and Send, or tap 🎤 Speak → say your line → tap again to send. Hit Coach my last answer any time.
📇 Live Client Mode
Enter what you know about a real upcoming appointment. The AI will play that client so you can rehearse before the real call. Use first names only — don't enter sensitive personal data.
Type your line and Send, or tap 🎤 Speak → say your line → tap again to send. Hit Coach my last answer any time.
📕 Playbook back-end
This is where you (the agency) control the training. Edit the scripts, the objections, and the best-practice model answers. The AI grades agents against exactly what you put here.
Coaching rubric weight
These are the five dimensions the AI scores. Names are editable; grading adapts.
Appointment-setting script
Aged / old lead script
Cleaning up an old file — ask for their help and re-explain the product. Feeds the coach and the dedicated aged-lead scenario.
Tonality & posture
Low-and-slow receptionist opening. The coach weights this heavily, especially early, and the AI client reacts to your tone.
Defusing an angry / hostile answer
Apologize for the bad timing, don't push, exit gracefully, retry later. Feeds the coach and the dedicated angry-answer scenario.
Presentation script
Application expectations
Bank-style identity verification, but ease into the sensitive info (DL/SSN) last.
Critical period / equity-protection presentation
The signature equity-protection frame. Feeds the coach and the dedicated role-play scenario.
Beneficiary pivot (unhealthy insured)
When the proposed insured only qualifies for guaranteed issue — pivot to insuring the healthy spouse/beneficiary. Feeds the coach and its own role-play scenario.
Reset pivot (already covered → advanced markets)
Turn an "already covered" no-sale into a Debt-Free Life or annuity lead for the advanced-markets SME. Feeds the coach and its own role-play scenario.
The emotional "why" (no why, no buy)
The heart of every call. The coach flags a shallow "why" as the top fix on every scenario, and this powers the emotional-discovery drill.
Methodology (NEPQ & non-negotiables)
The AI grades and rewrites through this lens across every trainer, role-play, and scorecard.
Roddy Jessup master class
The two conversations, the 10-sits standard, "establish the situation," one-best-policy/three-price-points, and the close. The coach grades against this.
The 5 buckets (diagnostics & mindset)
Diagnose show rate, close rate, premium, issue rate, and energy — plus the "catch the bus / negotiation" mindset.
Mentor coaching voice
How the AI coach should sound when giving feedback.
Products offered
Product fit — term vs. final expense (health & age)
The coach uses this to grade whether the agent matched the solution type to the client's health/age (full-payoff term for healthy/younger; equity-protection / final expense whole life for significant health issues or older age).
Carrier & underwriting quick reference
Used to coach product fit in Live Client Mode.
CRM text snippets
Lead-outreach snippets agents can copy. Reference only.
No-show reduction sequence
Client personas (for Full Role-Play)
One persona per line. These fill the persona dropdown; "Random" picks one for the agent. Add as many as you like — describe age, family, mortgage, and vibe.
Health scenarios (for Full Role-Play)
One health background per line. A random one is assigned to the client each session (in addition to a built-in set), so agents practice fact-finding across many conditions. Write them in second person, e.g. "You have Type 2 diabetes on insulin with neuropathy."
Objections & model answers
⚙️ Settings
Connect the app to Claude so the client can respond and coach you live.
Claude API connection
Your key is saved in this browser (this device only) so you don't re-enter it, and is never sent anywhere except directly to Anthropic. Get a key at console.anthropic.com.
If a model name ever errors, check current IDs at docs.claude.com/models.
🎙️ Voice conversation
The client speaks out loud, and you can reply with your voice. To talk: tap “🎤 Speak,” say your whole line, then tap the same button again to send it. The client answers out loud. (You can also just type and hit Send.) Works best in Chrome or Edge.
The app hand-picks the two most natural voices on this machine — one male, one female — and uses only those. On Microsoft Edge these are the studio-quality neural voices (the most human by far), so agents should train there.
About this prototype
This is a working single-file prototype of At Bats (MP + Living Benefits). Your API key and voice settings are saved in this browser on this device (use “Forget saved key” to clear them); Playbook edits still live in the tab session — use Export to keep them. A production build would add real agent logins, saved history, manager dashboards, and a server-side key so agents never handle one. See the accompanying build spec.