ShiftMate Voice Training Simulator

What would you like to train today?

Custom-built voice and chat simulations for the conversations your agents actually have — based on your campaign scripts, QA rubrics and compliance objectives.

Capability over credentials — training that proves itself before it ships.

Objection handling

Compliance disclosure

Voice + chat

Custom battle arena CLIENT CAMPAIGN / LIVE

Your agent

Active battle setup

Medium
Sean M.
VS

Campaign ABSA Enhanced FlexiFuneral

Scenario Skeptic — objections + benefit probe

Goal Listen, paraphrase, address the objection, then close with confidence.

Safe practice environment

Opponent personas

The ShiftMate Talent Value Chain

One product. Part of a connected talent system.

ShiftMate connects the entire journey — from attracting candidates to assessing potential, training capability, matching talent, placing candidates, and improving performance on the floor.

01Attract
02Assess
03TrainYou are here
04Match
05Place
06Perform

Voice Simulator prepares candidates for live calls using SA campaign scripts and personas — so agents arrive call-ready, not learning on your customers.

Works with:AI Interviews•Talent Sourcing•Assessments

Next Step

Match trained agents to open roles instantly

Explore Marketplace

ShiftMate Simulator Builder

How we build your roleplay simulator

From scenario setup to scoring logic, every simulator is configured to match the job, the customer, and the outcomes you care about.

The 7-step builder workflow

1

Set the scenario

Define the call type, campaign, product, and use case.

Call type

Inbound Sales

Campaign

FlexiFuneral Q2

Product

Premium Plan

Use case

Objection handling
2

Select the voice

Choose the voice style, accent, tone, and speaking pace.

Voice style

Professional

Accent

South African – EN

Tone

Warm / Friendly

Pace

Slow Fast
3

Add the brain

Plug in the AI reasoning layer, knowledge, objection handling, and logic.

Knowledge source

Product KB v2.3

Objection handling

Enabled

Reasoning depth

Balanced

Logic rules

24 active
4

Character persona

Define who the customer is, attitude, demographics, backstory, skepticism level, and behaviour.

Attitude

Cautious

Skepticism level

Medium-High

Behaviour

Asks probing questions

Persona

● ● ● +
5

Objectives

Specify what the trainee must achieve in the simulation.

Primary objective

Handle objections

Secondary objective

Close the sale

KPI focus

Conversion ▰▰▰
6

Rubric

Set the scoring framework, compliance checks, and coaching criteria.

Framework

Sales Excellence

Compliance checks

12 checks ›

Coaching criteria

15 criteria

Pass score

80% ●
7

Scripts

Add talk tracks, disclosures, prompts, and guardrails.

Talk tracks

18 items

Disclosures

6 items

Prompts

8 items

Guardrails

On ●

✦  All connected. Everything works together.

Your selections power a realistic, intelligent, and measurable roleplay experience.

Realistic customer behaviour

AI personas react naturally based on attitude, context, and conversation flow.

Configurable scoring

Score what matters with custom rubrics, compliance checks, and pass thresholds.

Voice + persona matched to the role

The right voice and persona create authentic practice for real-world conversations.

01 / Custom-built roleplay simulators

Your campaign.
Your rubric.
Your reality.

We don't hand over templates. We build simulators around your real scripts, QA scorecards, product rules and compliance objectives — so your team trains for the conversations that matter.

Pick the skill: opening, closing, recovery or objections.

Every persona has a profile, context, triggers and disposition.

Score against the behaviours your QA team already measures.

MTN

MTN | Training that connects

Scenario Studio

Built for real MTN customer conversations

Active
Full nameNomsa Dlamini
CityWelkom, Free State
Product interestData bundle upgrade
Current bundleMTN 10GB Anytime (30-day)
Monthly spendR149 – R199
Tenure10 months
Handset / deviceSamsung Galaxy A24 (4G)
Network frustrationSlow data in the evenings, signal drops
😟

Customer disposition

Confused about data pricing and what's included. Frustrated by slow speeds after 6PM and occasional signal drops at home. Compares bundles often, skeptical of upsells, and wants clear value. Responds best to calm reassurance transparency and practical recommendations.

Price sensitiveSeeks clarityCompares bundlesSkeptical of upsells
Aligned to MTN policy, POPIA and compliance requirements

02 / Live in-flight coaching

A second brain,
quietly in the room.

While an agent is on a real or training call, the coach watches for the exact QA triggers that matter. Suggestions appear only to the agent — zero distraction to the customer.

In-flight coach / active call
TRANSCRIPT · 00:04:38FAIS / FUNERAL
CUSTOMER“I already have cover through my work. Why would I need another policy?”
AGENT“That makes sense. Many people we speak to already have something in place. The difference is—”
CUSTOMER“So what does this cost me every month?”
AM

Listening for next response...

Private coaching channel — never visible to the customer.
Watch our AI coach live

Your scenario library

Real customers you'll encounter

Explore campaigns
Precious

WARM — BURIAL-SOCIETY COMPARISON

Precious

Warm

Budget-conscious admin clerk who already has a burial society.

Objective

Create a genuine need by framing the burial-society gap, confirm policyholder, confirm age before quoting.

Edwin

EDGE CASE — ANNUAL-INCREASE ACCURACY

Edwin

Skeptical

Johannesburg IT manager who holds one incorrect belief about the annual increase.

Objective

State the increase correctly per policy terms; do not lock in a figure before checking.

Ian

DRILL: OBJECTION CRAFT

Ian

Skeptical

objection-handling drill — price and accidental-benefit probe

Micro-drill

Objective

Handle the price-value objection with the three-step framework (listen, paraphrase, address) and explain the accidental-benefit clause accurately without over-promising.

03 / Synthetic voice lab

We use AI to train our AI.

ShiftMate runs AI agents against AI customer personas thousands of times before a human agent ever touches a scenario.

Every objection path is explored, every edge case caught, every persona stress-tested. The result is a partner that already knows every way a call can go wrong.

AI Agent
AI Persona
Score
Calibrate
repeat until sharp

04 / How the simulator thinks

Built on agents, loops, and graphs.

Every training session is not a linear script — it is a structured loop. A planner agent sets the objective. An executor agent runs the conversation. A checker agent scores the result with fresh context. The loop repeats until the agent clears the bar. Only then does it stop.

PLAN

Define what passing looks like

Before the call starts, the simulator sets the objective — which skills to test, which rubric to apply, which persona to use. The bar is fixed before the session begins.

EXECUTE → CHECK → ITERATE

Run the loop until the score clears

The conversation runs. A separate scoring agent — one that has never seen the agent's prior attempts — marks it against the rubric. Every gap is named exactly. The agent tries again. The loop runs until the score holds.

STOP

Ship only what cleared the bar

When the rubric passes, the session ends. Not when the trainee feels ready. Not when the coach thinks it looked good. When the loop says so.

The fresh context rule

The agent that ran your call cannot mark your call.

The scoring node always gets a fresh context. It has never seen your prior attempts, your warm-up session, or your practice runs. A checker that shares context with the thing it is checking is not a checker — it is the same process pretending to be two. Our rubric scorer is genuinely independent. That is what makes the score mean something.

The training graph

Planner node

Sets the objective

Executor node

Runs the conversation

Checker node

Fresh context — scores it

Iterator node

Loop or stop

loop until score clears · then stop

The same loop structure powers the Synthetic Voice Lab — which tests every version of our training before it ships to your simulator.

Ready for the floor

Make the next call the easy one.

See what a simulator built around your campaign looks like.