Know What You're Practicing — And How You'll Be Judged
The way a learner finds the right roleplay scenario, sees the rubric and pass threshold before they start, and reviews how their own past attempts actually went. "What should I practice before Thursday's escalation review?" "What is this scenario grading me on?" "Did my last de-escalation attempt pass?" — answered in chat. 4 read-only tools. The agent's RISKY_TOOLS list is empty.
Why Practice Gets Skipped Even When The Scenarios Exist
The roleplay engine isn't the hard part — Practice Hub already runs the AI persona, the rubric, and the scorecard. The friction is everything around the attempt: finding the right scenario, knowing the bar before you start, and remembering what went wrong last time. Practice Hub Agent closes that gap without a library crawl.
Nobody Knows Which Scenario To Practice
A rep has a renewal call Thursday and a library with 60 published scenarios. Which one rehearses the objection they're actually going to hit? Browsing categories takes longer than the practice itself, so the rep does neither and winds up rehearsing on the customer.
Learners Walk In Blind To The Rubric
You finish an attempt, get a 62, and have no idea why. The criteria were there the whole time — "acknowledged the frustration before problem-solving," "confirmed the resolution" — but nobody read them before the roleplay. Scoring feels arbitrary, and arbitrary scoring kills the willingness to try again.
The Same Mistake Repeats Because Nobody Rereads Their Own Attempts
The scorecard was specific and useful. It was also three weeks ago, buried under two more attempts, and nobody opens attempt history voluntarily. So the learner repeats the exact criterion they lost points on last time — with no memory that they lost points on it.
"Does This Actually Count Toward My Certification?" Is Unanswerable
Some scenarios are certifiable, most aren't. Community and curated scenarios are great practice but never mint a credential. Learners can't tell which is which from the tile, so they either over-invest in practice that doesn't count or skip the one that does.
Discovery Cost Exceeds Practice Cost
A de-escalation rehearsal is eight minutes. Finding the right de-escalation scenario, reading its setup, and confirming it matches the situation is twelve. When the overhead is bigger than the exercise, the exercise loses — every time, for every learner.
A Tool That Reports Your Fumbles Kills The Practice
Practice is the one place people are supposed to be allowed to be bad at something. The moment a learner suspects their fumbled attempts are visible to their manager through a chat assistant, they stop taking the hard scenario and start farming the easy one. The psychological safety is the product.
Practice Hub Agent At A Glance
AI Practice Hub
Find the scenario, understand the rubric, review your own scores.
Inside Practice Hub Agent — The Actual Capabilities
Every block below maps to a real tool against the Practice Hub app. Strictly read-only — the agent surfaces scenarios, rubrics, and the learner's own results, but it never starts a roleplay, never submits a turn, and never scores an attempt. The roleplay runs in the app UI; the score comes from the app's rubric engine.
Find The Right Scenario — By The Situation You're Facing
"What's there for de-escalation?" "Any cold-call practice?" — the agent matches your words against published scenario titles and categories and returns the shortlist with difficulty and whether the scenario can certify. Only published scenarios; drafts and other people's unpublished work never appear.
- search_scenarios — natural-language keyword matched against published scenario title and category.
- Certifiable flag surfaced up front — you know before you invest whether a pass can mint a credential.
- Difficulty on every result — beginner, intermediate, or advanced, so the rehearsal matches the stakes.
- Published only — the agent never surfaces draft or in-review scenarios it shouldn't.
Read The Rubric Before You Roleplay
For any scenario, the agent returns the full setup — the situation, the learner objective, the persona you'll be roleplaying against, the rubric criteria you'll be scored on, and the pass threshold you have to clear. Knowing the bar before you start is the single cheapest way to raise a first-attempt score.
- get_scenario_details — situation, learner objective, persona, rubric criteria, and pass threshold for one scenario.
- The persona is named — you know whether you're facing an angry customer, a skeptical VP, or a reacting audience.
- Criteria are visible, not a surprise — the rubric labels you'll be graded against, before the first turn.
- Read-only — the agent explains the scenario; the roleplay itself starts in the app.
Your Own Attempts — And Only Yours
"How did my last few attempts go?" — the agent returns your recent practice attempts with scenario title, status, overall score, and pass/fail. Self-only, deliberately: it cannot show anyone else's results, and there is no org-wide practice analytics tool.
- list_my_attempts — the asking user's own recent attempts with scenario title, status, overall score, and pass/fail.
- Strictly self-only — a manager cannot ask this agent for a team member's practice scores. There is no such tool.
- Repeat-attempt context — see the same scenario twice in your history and the improvement is right there.
- Read-only — reviewing history never alters it, and never re-scores an attempt.
How Scoring And Training Credit Actually Work
"Why did I get a 62?" "Does passing this earn me anything?" — the agent explains the scoring model: weighted rubric criteria scored 0–100, combined into an overall score, compared against the pass threshold, with delivery signals reported alongside. And which scenarios can count toward Training certification.
- explain_scoring — how the rubric, weighting, pass threshold, delivery signals, and Training credit work. No lookup needed.
- Certification path made explicit — only official, certifiable scenarios can count a passing attempt toward Training.
- Manager override explained — where enabled, a reviewed override becomes the final score, and learners know that up front.
- Static explanation, not a personal breakdown — this tool describes the model; your own scores come from your attempt history.
Outcomes Teams Can Measure
Practice Hub Agent's job is to cut the overhead around an attempt — finding it, understanding it, and learning from the last one. Compare against your pre-agent baseline.
- Scenario discovery time — minutes from "I need to rehearse X" to starting the right scenario, vs the pre-agent library crawl.
- Repeat-attempt rate — share of learners who attempt the same scenario more than once, the clearest signal practice is sticking.
- Score improvement first → latest — delta between a learner's first and most recent attempt on the same scenario.
- Rubric-reviewed-before-attempt share — share of attempts preceded by a scenario-detail lookup, the cheapest lever on first-attempt scores.
- Practice-to-certification conversion — share of passing attempts on certifiable scenarios that convert into a Training credential.
Intentionally Read-Only · The Roleplay And The Scoring Stay In The App
Practice Hub Agent's RISKY_TOOLS list is empty. It finds scenarios, explains rubrics, and reports the learner's own scores — it never starts a roleplay, never submits a turn, never scores an attempt, and never exposes another learner's results. Roleplay happens in the Practice Hub UI; scoring is the app's rubric engine.
- Zero write tools — RISKY_TOOLS list is empty. No attempts started, no turns submitted, no scores written or overridden.
- Self-only results, on purpose — no manager rollup, no team dashboard, no org-wide practice analytics tool exists. Practice is where people get to be bad at something first.
- Published-only discovery — search returns published scenarios; an unpublished scenario's content stays with its author.
- Audit trail on every retrieval — every read logs the requesting user, the tool used, and the parameters.
WHAT TEAMS TRY INSTEAD
The four alternatives — and why none of them know your rubric or your last attempt
When someone needs to rehearse a hard conversation, they reach for one of these four. None of them combine your organization's authored scenarios, the rubric you'll actually be scored against, and your own attempt history in one chat surface — with the results staying private to you.
Generic AI roleplay by prompting ChatGPT or Claude
"Pretend you're an angry customer" — improvised persona, invented grading
- Practice Hub Agent surfaces YOUR company's authored scenario — the real objection, the real policy, the real persona your peers encoded — not a generic archetype
- The rubric is the one your attempt is actually scored on, with the real pass threshold, not a grade the chatbot made up on the spot
- Your attempt history persists and improvement is measurable; a chat window forgets you practiced at all
Standalone sales or service roleplay vendors
Second login, second content library, second set of results
- Joins with Training and Skills & Certifications so a passing attempt on a certifiable scenario mints a real credential — vendor roleplay ends at a score
- No per-learner roleplay license on top of the platform your workforce already logs into every day
- Scenarios are peer-authored inside your tenant, so the practice reflects how your best reps actually handle it — not a vendor's stock library
Live roleplay with a manager or a peer
High quality, impossible to schedule, and never neutral
- Available at 11pm the night before the call, not whenever the manager has a free thirty minutes
- Consistent rubric scoring instead of one manager's mood — and no career-visibility anxiety about fumbling in front of your boss
- Frees live coaching time for the conversations that genuinely need a human, after the mechanics are already rehearsed
The manual fallback — skip practice, learn on real customers
The default when finding the scenario costs more than the practice
- Cuts discovery to one question, so the eight-minute rehearsal stops losing to twelve minutes of browsing
- The rubric is visible before the attempt, so the first try is a real attempt rather than a throwaway calibration run
- The mistake gets made against an AI persona at no cost, instead of against the renewal that was on the line
PLATFORM ADVANTAGE
Practice Hub Agent inherits everything the platform already runs
A standalone roleplay vendor has to build each of these. Practice Hub Agent gets them for free.
Cross-app data plane
Scenarios link to Training courses and Skills & Certifications credentials, so "does this count?" has a real answer instead of a guess.
Self-scoped reads
Attempt lookups are hard-scoped to the asking user server-side. There is no parameter, prompt, or phrasing that returns another learner's practice results.
Audit trail & retention
Even read-only lookups log to AiApiLog with the requesting user — useful when L&D compliance or certification audit questions come around.
Translation in 100+ languages
Frontline learners find scenarios and read rubrics in their working language, matching the multi-language roleplay the app already supports.
Frontline-ready on a phone
A rep finds and reads the scenario on the way to the site visit — same mobile app, no separate learning tool to install.
RubyLLM-grounded model tiering
Scenario lookups run cheap; rubric explanation and attempt review route up. Automatic per call.
INDUSTRY FIT
Industries where rehearsal AI moves the most weight
Practice Hub Agent shines wherever the hard conversation is the job — and getting it wrong the first time is expensive.
Retail / Hospitality
Service recovery and de-escalation rehearsed before the shift, by associates who find the right scenario in one question instead of skipping practice entirely.
Healthcare
Difficult family conversations and patient-handoff scripts rehearsed privately, with certifiable scenarios feeding real competency records.
Financial Services
Advisors rehearse disclosure conversations and objection handling against a rubric that encodes what compliance actually requires.
Manufacturing
Safety stand-downs, incident conversations, and shift-lead coaching rehearsed by supervisors who never got formal management training.
Technology
Cold calls, renewal saves, and escalation calls rehearsed against peer-authored scenarios that carry the real objections reps hit this quarter.
Public Sector
Citizen-facing de-escalation and intake conversations rehearsed with rubric consistency across offices, inside the tenant's own data boundary.
WHY MANGOAPPS WINS
An embedded practice agent beats a roleplay vendor, a horizontal chatbot, or a manager's calendar on every axis
The argument learners, L&D, and IT all share — and the one a standalone roleplay vendor structurally cannot answer.
Cheaper than the alternatives
No per-learner roleplay-vendor license, no per-seat ChatGPT license, and no manager hours burned running live roleplay sessions that an AI persona can carry.
More secure
Read-only by design. Attempt results hard-scoped to the asking user. Published-only scenario discovery. Every retrieval logged. Nothing leaves the tenant.
Easier to deploy
Already deployed if Practice Hub is enabled. Turn the agent on and learners can find scenarios by description the same day.
Easier to use
One chat surface for finding a scenario, reading its rubric, and reviewing your own scores — no library crawl, no separate learning portal.
Easier to manage
Scenario lifecycle, certifiable flags, and pass thresholds live in the same admin console as every other app. One audit log, one access model.
Easier to extend
New read capabilities ship as agent tools against the same scenario and attempt records — no vendor roadmap to wait on, no content library to re-import.
AI is actually better
A generic chatbot can improvise an angry customer. Only Practice Hub Agent knows which of your authored scenarios matches the call you have Thursday, what rubric it scores you against, what threshold you have to clear, and how your last attempt at it went.
At our head office within HR, marketing, and finance, we use ChatGPT quite a bit, and MangoApps AI Helper is a useful tool for us. As I go around to the stores, I’ve been showing them how the AI Intranet Helper can find things and answer plain language questions about our resources, and they’re quite excited about that.
Alison van Zijl
Head of Human Resources
Full House
Customer Success
MangoApps Customers
Organizations that trust MangoApps to run every workflow across their workforce.
Customer Case Studies
Customer Case Studies
Customer Case Studies
Customer Case Studies
Customer Case Studies
Customer Case Studies
Frequently Asked Questions About Practice Hub Agent
4 read-only tools — search_scenarios to find published scenarios by keyword or topic, get_scenario_details for one scenario's situation, objective, persona, rubric criteria, and pass threshold, list_my_attempts for the asking user's own recent attempts and scores, and explain_scoring for how attempts are scored and how practice counts toward Training.
No. RISKY_TOOLS is empty — Practice Hub Agent is strictly read-only. It never starts an attempt, never submits a turn, and never scores or re-scores anything. The roleplay runs in the Practice Hub app UI, and the score comes from the app's rubric engine. The agent points you at the scenario; you practice in the app.
No. list_my_attempts returns only the asking user's own attempts, and there is no org-wide practice analytics tool on this agent — no team rollup, no coaching dashboard. That is deliberate: practice is where people are allowed to be bad at something, and a tool that reported your fumbled attempts to your manager would kill the practice.
Published scenarios only. Drafts and in-review scenarios stay with their author — the agent applies the same visibility gate as the web UI, so it never leaks unpublished scenario content to other members.
explain_scoring returns a static explanation of the scoring model — weighted rubric criteria, the overall score, the pass threshold, delivery signals, and Training credit. It is not a personalized breakdown of one attempt. Your own scores come from list_my_attempts, and the full evidence-backed scorecard lives in the app.
Scenario discovery time, repeat-attempt rate, score improvement between a learner's first and latest attempt on the same scenario, share of attempts preceded by a rubric lookup, and practice-to-certification conversion. Compare against your pre-agent baseline.
Let's Talk
Since 2008, we've been building the employee platform for the frontline, earning the trust of 2 million+ users and an NPS of 78.
Why Choose Us?
- AI-Ready Platform: One intelligent place for every employee and workflow.
- Top Security: HITRUST, ISO & SOC 2 certified.
- Exceptional UX: Delightful on mobile and desktop.
- Proven Results: 98% customer retention rate.
Trusted by Legendary Companies:
Prefer to explore first? Ask AI about Practice Hub AI Agent →