Practice Hub

Walkthrough

Apps & Agents Directory

Practice Hub

Peer-authored AI roleplay — employees rehearse real conversations and presentations, with coach-style scored feedback.

MangoApps

Book a walkthrough Ask AI Download datasheet

Overview

Practice Hub is the practice layer for your workforce, completing the Learn → Practice → Do loop alongside Training. Employees rehearse real workplace situations — de-escalation, difficult 1:1s, interviews, service recovery, harassment-response, even presentations and pitches — against an AI that plays the other person (or the audience), then get an AI-coached scorecard with concrete, criterion-level feedback and delivery metrics. Most scenarios are authored by peers (your best reps, nurses, and shift leads encoding how they actually handle things), curated by L&D, and promoted to official for certification. Behavioral compliance and soft skills both live here; knowledge checks stay in Training.

Learning Growth

Highlights

Rehearse real workplace moments — a tough 1:1, a de-escalation, an interview, a service recovery — against an AI that plays the other person, by voice or by text.

Practice presentations, pitches, and speeches in Delivery mode, where the learner holds the floor and the AI plays a reacting audience that asks follow-up questions.

Peer-authored first: your best reps, nurses, and shift leads encode how they actually handle a situation, so practice reflects reality — not an L&D deck.

Every attempt ends with an AI-coached scorecard: median-of-N rubric scores backed by the learner's own quoted lines, delivery metrics (pace, filler words, talk-to-listen), and narrative coaching.

A community → curated → official lifecycle lets L&D promote the best peer scenarios; passing an official scenario mints a real Skills & Certifications credential, not just a completion log.

Optionally reward peer authors with Recognition points on promotion, upload a real call or meeting recording for private analysis, or turn a weak criterion into a focused micro-drill.

Capabilities

Practice Modalities & Runtime

Voice roleplay — speak your reply and hear the AI persona answer back in its own voice
Live speech-to-speech roleplay — a fully spoken conversation (admin-gated, uses voice minutes)
Conversation vs Delivery interaction mode (spar with a counterpart, or hold the floor to an AI audience)
Multi-persona panels — an interview panel or buying committee where the AI picks who replies next
Interview rehearsal — run your own Interview Scheduler question set against an AI candidate, then get scored
Per-scenario modality control (learner's choice / voice-only / text-only)
Multi-language roleplay — persona and feedback in the learner's language

Scenario Authoring & Curation

Peer-authored scenarios (situation, objective, personas, weighted rubric, tags)
AI-assisted scenario drafting from a plain-language description
Configurable AI personas (role, temperament, goals, playback voice)
AI quality-gate pre-screen (safety, PII, off-brand) with author recheck
Company context pack — products, objections, policies, and tone that ground every persona and scorecard in your company's reality
Turn a real recorded meeting (AI Notepad transcript) into a draft rehearsal scenario
Community → curated → official lifecycle with single and bulk promotion
Peer ratings (1–5) feeding curation ranking, plus a system starter-template gallery

AI Scoring & Coaching

Median-of-N rubric scoring at temperature 0 (outlier-resistant)
Evidence-first scoring — the learner's own quoted lines under each criterion
Delivery metrics: filler words, word count, talk-to-listen ratio
Speech pace (words per minute) from measured voice-turn duration Computed only over voice turns; typed turns never fabricate a pace
Narrative feedback with a pass/fail scorecard and manager override
"Show me a 95" exemplar and coach chat on the learner's own scorecard

Personal Progress & Coaching Memory

Rolling coach memory you can see, correct, and set a goal on
Personal skill radar — per-competency scores across your recent practice, weakest area first
Delivery trend over time — pace, filler rate, and talk-to-listen tracked against conversational benchmarks
Recommended for you — next scenarios picked from your own weak spots, each with a stated reason

Skills, Certification & Assignments

Passing an official, certifiable scenario mints a real Skills & Certifications credential
Recertification cadence — expiry and renewal by re-passing the scenario
Write-back to Training as a completion record on pass
Skill-linked targeting — gap → practice recommendations via the Workforce Readiness Graph
Assign to users or departments with due dates and nudges
Manager review queue with per-criterion score override

Recording Analysis, Drills & Incentives

Upload a real call, meeting, or rehearsal recording (audio or video) for private AI analysis
Analyze an AI Notepad meeting transcript against a rubric — no upload needed
Private by default — share a specific analysis with your manager only when you choose
Consent attestation on upload and auto-delete of raw files after a retention window
Micro-drills — turn a weak scorecard criterion into a few focused exchanges
Creator incentives — Recognition points awarded to authors on promotion (opt-in)

Admin & Analytics

Adoption, pass-rate, and most-practiced scenario dashboards
Assignment funnel and shelf-decay + department cohort views
Team competency rollup from per-criterion scorecard scores
Weekly practice plan (dashboard card, plus an opt-in Monday Inbox digest)
Personalize coaching with workplace signals (review themes, recognitions, skills) — admin-toggled

Limits & Specs

Max recording upload: 500 MB (audio or video)
Recording file retention: 30 days default (admin-configurable); transcripts and scorecards are kept
Scoring method: Median of 3 independent evaluations at temperature 0
Certification level granted: Proficiency 3 (demonstration-level)
Pricing: Requires a license — opt-in per tenant

Use cases

De-escalation & service recovery

A frontline associate practices calming an upset customer against an AI persona and reviews their scorecard before facing the real thing.

Behavioral-compliance readiness

Instead of clicking through a harassment-response slide deck, an employee demonstrates the right response in a roleplay — and a pass on an official scenario mints a real certification.

Present with confidence

Before a town hall or a customer pitch, an employee rehearses in Delivery mode against an AI audience and gets coached on structure, clarity, and handling questions.

Difficult 1:1s and feedback

A new manager practices delivering hard feedback in a simulated 1:1, then asks the coach chat what to try differently next time.

Peer knowledge capture

A top performer authors a scenario encoding how they handle a common objection; when L&D promotes it, they earn Recognition points for turning tacit know-how into team practice.

Learn from a real call

With admin consent enabled, an employee uploads a recording of a real sales or support call and gets a private, rubric-scored breakdown of their own performance.

Resources

Practice Hub Datasheet (PDF)

Datasheet

Features, capabilities, pricing model, and FAQ on one downloadable sheet — share it with your team.

Getting started with Practice Hub

guide

Enable the app, load the sample scenarios, and run your first roleplay end to end.

Authoring a great practice scenario

guide

Write the situation, configure the AI persona(s), and build a weighted scoring rubric.

Curating peer scenarios (community → curated → official)

guide

How L&D reviews the AI quality-gate results and promotes the best peer scenarios — and which can certify.

Learn → Practice → Do: closing the training loop

blog

Why rehearsal beats click-through completion, and how Practice Hub feeds Training's certification spine.

How practice scoring works

help

Rubric scoring, the pass threshold, delivery metrics, and manager review/override.

FAQ

Peers are the primary authors. Any employee can draft a community scenario with AI assistance; L&D curates the best and promotes them to official. Only official scenarios can certify.

Training does Learn — watch a lesson, pass a knowledge quiz. Practice Hub does Practice — rehearse the behavior under pressure and get scored. Passing an official, certifiable scenario mints a real Skills & Certifications credential and records a completion in Training, rather than replacing them.

Yes. You can speak your replies — they're transcribed and the AI persona answers back in its own voice — and an admin can turn on live speech-to-speech roleplay for a fully spoken conversation. Text is always available as an alternative, and there is no webcam or facial analysis.

Delivery mode is for presentations, pitches, and speeches: the learner holds the floor while the AI plays a reacting audience that asks follow-up questions. Scoring judges structure, clarity, audience awareness, and how you handle questions — long uninterrupted turns are expected, not penalized.

If your admin enables recording uploads, you can upload a real call, meeting, or rehearsal (audio or video) and get it transcribed and scored against a scenario's rubric. These analyses are private to you, require a consent attestation, and the raw file auto-deletes after the retention window (30 days by default; the transcript and scorecard are kept).

For a scenario marked official and certifiable, a pass writes a real Skills & Certifications credential — an EmployeeSkill holding plus an EmployeeCertification period at demonstration-level proficiency — for each catalog skill the scenario develops that requires certification. It runs the same certification path as the admin form, batch assign, and CSV import, so expiry (via the scenario's recertification interval), renewal, and the verification queue all behave identically. The pass is also recorded in Training as an external completion. Practice Hub doesn't keep its own separate credential store — it feeds the one Skills & Certifications ledger.

Yes. An interviewer can turn an existing Interview Scheduler question set into a private practice scenario and run their own questions against an AI candidate, then get the standard scored review — covering question delivery, active listening and follow-ups, candidate rapport, and staying within legal guardrails. The draft is the interviewer's own and never enters the shared library unless they publish it through the normal curation gates.

The coach keeps a rolling memory of each learner's focus areas and strengths that the learner can view, correct (remove an item), and set a personal goal on — removed items may be re-observed if the behavior reappears. Admins can optionally let coaching draw on workplace signals like performance-review themes, recognitions, and skills gaps; that whole signal layer is a single admin toggle and its influence is surfaced in the learner-visible coach profile.

Live speech-to-speech roleplay is off by default and must be enabled by an admin, because each live session spends real voice minutes from your tenant's balance and is metered per session. Everyday practice — typing, or speaking a reply and hearing the AI answer back — doesn't require it.