For the cost of one classroom day: 7 hours of practice per person, instead of one hour at best. See the maths

Train and assess Practise with no audience, improve for real

Role-plays with an AI conversational agent, to learn and assess skills.

From their own computer, your teams talk out loud with an agent playing the customer, the team member or the candidate. They try again as often as they like and walk into the real conversation with confidence. And you see who improves.

Book a demo
No need to change your LMSHosted in FranceFlat fee per user
Role-play session: on screen, the conversational agent plays a head of procurement; the employee practises on webcam Head of procurement 04:12
A manager practising from her computerClaire
Head of procurement · AI agentYou announce +8% when your service has got worse. Why should I stay?
Claire, manager · speaking That’s a fair point. Let’s go through the two incidents from March together.
Listening and rephrasing+12 points since session 1
Illustrative example, fictitious data

Works alongside the LMS you already use

  • Moodle
  • 360Learning
  • Docebo
  • TalentLMS
  • Cornerstone
  • Rise Up

How it works

Fifteen minutes, three steps.

Nothing to install, no trainer to book: a browser, a microphone and the will to improve.

  1. Catalogue of ready-to-use Apprentisia scenarios1 min

    I pick my situation

    A negotiation, an annual review, an unhappy customer: taken from your line of work, or from our catalogue.

  2. Voice conversation on a video call with a conversational agent10 min

    I speak, it answers

    Out loud, as in a video call. The agent reacts, objects and pushes back, with the personality it has been given.

  3. Assessment grid with weighted criteriaright away

    I know where to improve

    Feedback on each criterion, the sentences that worked and the ones to rework. And I go again whenever I like.

Why we exist

Nobody learns to swim by reading a book.

The conversations that matter rarely happen twice. Apprentisia gives people a place to rehearse them, with no one watching and nothing at stake.

Role-play ties up a manager, happens rarely and is never measured: so people end up learning on real customers.

Before

People practise on real customers

Every botched call, every badly handled negotiation, every misjudged candidate is paid for in the field.

During

Everyone rehearses as often as they like

A conversational agent plays the customer, the colleague or the member of the public. No audience, no consequences, as many attempts as needed.

After

You finally see who is improving

A report for every session, criteria set in advance, progress tracked over time.

First use

Train through practice

Practise in realistic conditions, through AI role-plays. Fifteen minutes between two meetings is enough.

  • Situations taken from your line of work
  • Difficulty that adapts to the level
  • Memory from one session to the next
Discover training

Second use

Assess skills

Objectively assess soft skills and hard skills, through tailored scenarios and role-plays.

  • Weighted criteria set in advance
  • Every score linked to a specific moment
  • Human review before anything is shared
Objection handling88
Listening and rephrasing74
Defending value91
Reviewed by a managerTime-stamped quotes
Discover assessment

Who is it for?

One engine, three uses, all very practical.

Sales teams

Your sales team faces objections before meeting the real customer.

Cold calls, price objections, negotiation: they replay the situation as often as needed, with a score every time and no manager tied up.

  • Scenarios based on your offers
  • Objections calibrated by level
  • Progress tracked by team
Train my sales team

Customer service

Your advisers keep their cool with angry customers.

Complaints, aggressive callers, bad news to deliver: they rehearse before the real call, and time-stamped quotes feed the debrief.

  • Personas with adjustable temperament
  • Time-stamped quotes for the debrief
  • Progress tracked by team
Train my advisers

Managers

Your managers rehearse difficult conversations.

Annual reviews, reframing, feedback, difficult announcements: they practise without risk and go again until they are ready.

  • Bespoke management scenarios
  • Measurement of listening and assertiveness
  • Tracking by team
Train my managers

After every session

Feedback that helps people grow, not a verdict.

The product isn’t the conversation. It’s the report.

The conversation is impressive in a demo. What holds up in front of a committee is the document it produces.

A score per criterion, backed by timestamped transcript excerpts, areas for improvement and human review. Exportable as PDF or to your LMS.

Karim B. · Annual negotiation · session 4 of 6Assessment report
87+9 vs session 3
Objection handling88
Listening and rephrasing74
Defending value91
Keeping control of the negotiation79
“That’s a fair point. Let’s go back over the two incidents in March and what we have changed since.”Transcript excerpt used to justify the objection handling score · 04:12
Confidence index 0.92Reviewed by a managerTimestampedPDF & xAPI export

Our method

An assessment that stands up when challenged.

A score nobody can explain is worthless the day someone challenges it.

Grounded in validated research

Our scoring grids rely on published frameworks, not on a hunch written into a prompt.

Never a single signal

What people say, how they behave in the situation, consistency across sessions: every conclusion cross-checks all three.

Every point is evidenced

The report links to the timestamped transcript behind each score. You can replay the moment and decide for yourself.

Memory from one session to the next

The AI measures real progress rather than a one-off performance, and keeps pushing on what is holding people back.

Adaptive difficulty

The persona toughens or softens its objections according to the level observed.

A human stays in control

A manager or trainer reviews and can correct every assessment before it is shared.

What we offer

First they practise. Then we assess them.

One engine, three ways to use it.

7 hours of practice for the price of one classroom day

Your employees practise on situations drawn from your business. You choose the plan based on the amount of practice you want.

  • One hour included per month
  • Progress memory
  • LMS integration

€29 to €59 excl. VAT per user per month on an annual commitment, one to ten hours included; 10% off from 50 users, 15% off from 200, 20% off from 500. Schools, universities and apprenticeship centres: from €5 excl. VAT per learner per month.

Off-the-shelf assessments, at a fixed price

Ready-to-use assessments for your recruitment and skills reviews.

  • Pre-screening
  • English (CEFR)
  • DISC and Big Five profiles

€29 excl. VAT per assessment taken or per pre-screened candidate, €24 from 50, €19 from 200. Credits valid for 12 months.

Bespoke, at a known price

We build the scenario, persona and scoring grid specific to your business, together with you.

  • Scoping workshop
  • Custom scoring grid
  • Your own content

€750 to €2,500 excl. VAT per scenario depending on complexity, or an authoring licence at €199 excl. VAT per month, support included, to build them yourself.

What will it cost you?

On the Essential plan, an employee who practises all year costs €348 excl. VAT, less than a single deal lost for lack of preparation. Estimate your budget based on your usage.

Before committing for a full year, a three-month pilot at €3,500 excl. VAT, fully deducted from your annual contract if you sign.

Your scenarios
Estimated annual budget €15,200 Excl. VAT, i.e. €19 per assessment taken at this volume (discounts from 50 and 200 attempts) Get in touch

Pricing: Essential at €29, Pro at €39 and Expert at €59 excl. VAT per user per month on an annual commitment (one, two and up to ten hours included), 10% off from 50 users, 15% off from 200, 20% off from 500; hours pooled across the users of the contract, the administrator is alerted before the total is reached and, if needs durably exceed the plan, an upgrade to the next plan is offered, never billed without agreement. Assessments and pre-screening: €29 excl. VAT per attempt, €24 from 50, €19 from 200. 20-minute sessions.

For your organisation

You describe the situation, the platform does the rest.

No code, no filming, no actors. You describe the situation, the platform does the rest.

A scenario in four steps.

  1. STEP 01Describe the situation

    The context, what is at stake, what the employee needs to achieve.

  2. STEP 02Choose the persona

    Off the shelf or bespoke: personality, objections, level of demand, language.

  3. STEP 03Set the criteria

    3 to 7 weighted criteria, each with its pass mark.

  4. STEP 04Open access

    Via a direct link or inside your LMS. Your teams can start practising the same day.

A learning path built like a diagram

Each step is a block: questions to ask, media to display, a document to upload. You connect the blocks; the AI suggests and refines.

  • Add a step in one click
  • Pass criteria assessed by the AI
  • Media and questions for each step
Apprentisia learning path editor: connected steps and the configuration panel for one step
One interface to run it all: scenarios, personas, reports.

Standards & LMS

Compatible with your LMS, no migration needed.

You keep your platform and your accounts. No LMS? A simple link is all you need.

Tested LMSs

  • Moodle
  • 360Learning
  • Docebo
  • TalentLMS
  • Cornerstone
  • Rise Up

Standards and connectors

  • SCORM 1.2
  • SCORM 2004
  • xAPI / Tin Can
  • cmi5
  • REST API (on request)
  • SSO SAML 2.0 / OIDC (on request)
How Apprentisia integrates with your LMS

A framework you can trust

AI runs the rehearsal. People decide.

Your employees practise in a safe space. Your data stays in France. And no assessment is shared without a human review.

No audience

You can get it wrong, hesitate and start again. Nobody is looking over your shoulder.

A human stays in control

A manager or trainer reviews, corrects or rejects every assessment.

Data hosted in France

Application and data in Paris, GDPR compliant, configurable retention periods.

0

trainers needed: practice runs on its own

15 min

for a typical session, between two meetings

4 wks

from scoping to the first scenario in production

×7

more practice per person, for the cost of one classroom day

Your data hosted in France.

GDPR, human oversight, full traceability: the compliance file is ready before your legal department asks for it.

Hosted in France

Application and data hosted in Paris, with the option of 100% Mistral AI. Providers and transfers outside the EU are listed and legally governed.

Pre-filled DPIA

You receive the GDPR data protection impact assessment already drafted, ready to adapt rather than write.

Human oversight

No automated decisions: a manager can approve, amend or overturn every assessment.

Data minimisation

Only what the assessment needs is kept, with configurable retention periods and deletion on request.

Six-monthly bias audit

Regular checks for scoring gaps between groups, with a report and corrective measures.

Decision log

Who assessed, when, against which grid, with which corrections. Exportable for an auditor.

How we compare

The market leaders are American and Israeli. We are sovereign.

Comparison of Apprentisia, Yoodli and Second Nature
CriterionYoodliSeattle · approx. $61.5M raised, $40M Series B in December 2025Second NatureTel Aviv / US · $38M raised, $22M Series B in October 2025ApprentisiaFrance · sovereign
Pricing model€25 to €50 per user per month€30 to €40 per user per month€29 to €59 per user per month, one to ten hours included, public prices
Documented scientific basis for scoringNoNoYes
Memory across sessionsNoYesYes
SCORM / xAPI exportNoYesYes
Deployment under your own brandNoNoYes
EU hosting and governanceNoNoYes

Comparison compiled on 3 August 2026 from public pricing pages and the press releases cited. Information subject to change. Sources: Yoodli, public pricing, Second Nature, pricing FAQ, Yoodli, $40M Series B (2 December 2025), Second Nature, $22M Series B (17 October 2025).

About us

Two founders, four partners, four areas of expertise working as one.

Lionel Drouin

Founder & CEO, EdTech & AI

ENSEIRB engineer with twenty years leading digital businesses. Former CDO of Visiativ (1,000 employees) and Fiducial.

  • Digital learning
  • SaaS
  • B2B scaling
  • Growth

Stéphane Pétrot

Chief Technology Officer

SaaS architect with fifteen years running platforms in production. In charge of platform security and compliance.

  • SaaS architecture
  • Ruby on Rails
  • Rapid MVPs
  • Tech recruitment

Thomas

Chief Commercial Officer

Business development backed by expertise in cyber security, GDPR and the EU AI Act: he sells compliance obligations that are achievable, built into the platform rather than left in a PDF.

  • Business development
  • Partnerships
  • EU AI Act
  • GDPR

Mickael

AI Expert

Digital transformation at ALTEN, in Toulouse and Montreal. Certified Scrum Product Owner.

  • Artificial intelligence
  • Technical development
  • Product Owner
  • Digital transformation
Lionel Drouin, co-founder of Apprentisia

What we believe

Nobody ever learnt to negotiate by listening to someone explain how to negotiate. You learn by getting stuck in, getting it wrong and trying again. That is why we founded Apprentisia.
Lionel Drouin · Co-founder of Apprentisia

Partners & resellers

Do you sell training? Build in our engine.

Training providers, schools, LMS integrators: Apprentisia can be deployed under your brand, with a distribution discount on the public price list.

Become a partner
  • Distribution discount. 30% off the public per-user price list (15% for education), white label as an option: you set your own resale price.
  • Full white label. Your name, your colours, your domain. Apprentisia never appears on the user side.
  • No going around you. We never approach the clients you bring us. A contractual commitment.
  • Compliance handled by us. Under your brand, the EU AI Act makes you the provider: we hand you the compliance file, already prepared.

Apprentisia, in brief.

What it is
A French online platform for role-plays with a conversational AI agent, by voice or in writing.
The two uses
Training through practice, and objective assessment of skills, both soft skills and hard skills.
Who it is for
Companies (L&D, HR, sales leadership), training providers, schools, universities and apprenticeship centres.
Pricing
Plans from €29 to €59 excl. VAT per user per month on an annual commitment; assessments from €19 excl. VAT; 3-month pilot at €3,500 excl. VAT.
Hosting
Application and data in Paris, GDPR compliant, human oversight of every assessment.
Integration
SCORM 1.2 and 2004, xAPI, cmi5: plugs into your existing LMS, or works from a simple link.

Frequently asked questions

What we get asked before signing.

All questions

Do we need to change our LMS?

No. The module exports to SCORM, xAPI or cmi5 and works alongside the LMS you already use. Without an LMS, a simple link is all you need.

How exactly are we billed?

Three plans per user per month, on an annual commitment: Essential at €29 excl. VAT (one hour included, about three role-plays), Pro at €39 excl. VAT (two hours), Expert at €59 excl. VAT (up to ten hours). Without commitment: €35, €47 and €69 excl. VAT. 10% off from 50 users, 15% off from 200, 20% off from 500. Hours are pooled across the users of a contract; the administrator is alerted before the total is reached and, if needs durably exceed the plan, an upgrade to the next plan is offered, never billed without agreement. Assessments and pre-screening: €29 excl. VAT per attempt, €24 from 50, €19 from 200. Schools, universities and apprenticeship centres: from €5 excl. VAT per learner per month. White-label resellers have a separate price list, billed per minute.

How long does it take to get a first scenario?

Around four weeks from the scoping workshop to going live, reviews included.

How is the score calculated, and can it be challenged?

Each criterion is defined in writing and weighted before the session. The report links every point to the timestamped transcript, and a manager can amend or overturn the assessment.

Where is the data hosted, and for how long?

In Paris, for both the application and the database. Some providers are American: they are listed, along with where the data is stored, on the Privacy & sovereignty page. Configurable retention periods, deletion on request.

Does the AI replace the trainer or the manager?

No. The AI runs the practice and measures it. Diagnosis, decisions and coaching remain human.

Which languages does the simulation support?

French and English in production, with other languages available on request. The conversation and the report can be in different languages.

Thirty minutes, no commitment

Thirty minutes on a real situation from your business.

We take a real case from your business, turn it into a scenario, and you try it out. No slides.

Pick a time slot

Opens the Calendly booking page (United States) in a new tab.

  • < 24 hResponse time
  • 30 minDemo length
  • 0 slidesNo sales pitch

Or write to us

No commitment · A reply from a real person within 24 hours