TTrainexus
Products
Dialogue Simulation PromptStudio Video Analysis
Trainexus
Scenarios Pricing For decision-makers Who it's for News FAQ Request trial access
AI-powered conversation simulation

Conversation practice with AI avatars.

Rehearse the conversation that matters before it happens: exams, job interviews, patient consultations. With a conversation partner who feels real.

Free trial access · GDPR-compliant · Hosted in Germany
The problem

You know it. But can you say it?

You had no one to practise with.
Once was not enough.
The first real attempt was the real one.
The solution

A conversation partner who listens and responds.

The avatar looks at you — eye contact, facial expression.
The avatar does not wait — silences and follow-up questions, just like the real thing.
The avatar probes — like a good examiner or a sharp recruiter.
You get feedback — specific, structured, comparable.
How Trainexus fits together

From scenario design to learner reflection.

Six stages, one seamless process. Explore the whole journey below.

The Dialogue Simulation is the core app. PromptStudio creates the scenarios. Video Analysis is an optional add-on module.

1The teaching idea

Start with the situation.What makes up a scenario?
Every scenario is described along four axes: role, behaviour, reaction and sequence. This makes the AI partner both controllable and realistic — probing, emotional or concise, as the situation requires.

You — or a group of learners — want to practise a specific conversation: an oral exam, a consultation with a patient or client, a conversation between a manager and a team member.

View example scenarios →
A learner considering which conversation to practise
2In PromptStudio

The idea becomes a prompt.

In PromptStudio you turn the teaching idea into something the language model can work with — role, behaviour, learning objectives. No technical background needed, just plain language.

More about PromptStudio →
3Practising with the avatar

Then you practise — in real time.How a conversation runs
Log in, pick a scenario, start speaking — every word is transcribed live, and the assessment follows at the end. A full run often takes just a few minutes, and you can repeat it as often as you like.

Learners speak with an avatar presented as male or female. It behaves exactly as defined in the prompt: probing questions, emotional reactions, vague statements or extra detail.

More about the Dialogue Simulation →
4Assessing what was said

Feedback on every statement.Two kinds of feedback
Formative attached to individual statements (inline, for revising) and summative as an overall assessment at the end. You see both the detail and the bigger picture — exportable as HTML and PDF.

After the conversation, learners receive feedback on each individual statement (inline) plus a concise overall assessment — as a dashboard and as a PDF.

Inline-Feedback zu einzelnen Aussagen
5Video Analysis — optional

How you come across, too.

Optionally, you can analyse non-verbal communication as well: voice, posture, gaze and facial expression. That is Video Analysis — an optional add-on module.

More about Video Analysis →
Annotated video: posture, gaze and gesture being analysed
6Reflection

What do you take away?

At the end you reflect: What did I learn? What can I do now? What do I want to work on next? You record these thoughts in the scenario feedback.

Reflection and feedback on the completed scenario
Scenarios

Practise the conversation before it counts.

From oral exams to difficult conversations — 55 scenarios in the public catalogue, more than 300 in live use; new ones are built on request.

And more than 40 others in the catalogue.

From "conversation with a short-tempered professor" through nonviolent communication and return-to-work talks to medical licensing exams — searchable and filterable.

Explore the full catalogue

Secure. Under your control. Data-minimising.

Accounts, conversations and feedback are stored in Germany (Hetzner). AI services are covered by data processing agreements and EU standard contractual clauses; transmitted data is never used to train AI models. Details & data flow →

§

GDPR-compliant

Active deletion after every session (with an audit log) and pseudonymous account identifiers. Voice and facial expression are only analysed if you explicitly consent beforehand.

DE

Hosted in Germany

Database at Hetzner (Germany). AI and streaming services (OpenAI, ElevenLabs, HeyGen) under DPAs and SCCs; Video Analysis runs entirely in Germany (IONOS, Frankfurt) — no video ever leaves Germany.

Data-minimizing techniques

No recording on the avatar side, no text retention in text-to-speech, and raw Video Analysis footage is deleted immediately after processing.

Rooted in research

Developed at the University of Bayreuth — now used by organisations well beyond it.

1.911
active users
666 h
of conversation training (cumulative)
300+
scenarios in live use
Scientifically accompanied: a study with 390 students on exam preparation (University of Bayreuth, publication in preparation). The same platform runs white-labelled — among others as the exam simulator of ZHL Bayreuth and as "Coach Kilian" for the German tax adviser exam.
University of Bayreuth Centre for University Teaching (ZHL)
"The best preparation for a conversation is to practise a conversation."

Who we build for

Universities & research

Exam simulation for degree programmes, OSCE preparation, Moodle LTI integration, departmental pilot projects.

Schools & vocational colleges

Colloquium and school-leaving exam preparation, interview training, speaking practice — safe and without embarrassment.

Companies & corporations

Compliance training at team scale, leadership and sales conversations, scalable by link — without a per-head trainer budget.

Consulting & Freelancers

Client and patient consultations, professional body and association training, white-label solutions.

Deciding whether to roll this out at your institution? To the decision-makers' page — data protection, integration & procurement →

The team

A startup rooted at the University of Bayreuth — teaching expertise, research and engineering under one roof.

Paul Dölle

Paul Dölle

Teaching design and management
Dr Carl-Christian Fey

Dr Carl-Christian Fey

Engineering
Tobias Hallmen

Tobias Hallmen

Engineering and conversation research

Frequently asked questions

What is a conversation simulation?

A realistic practice conversation with an AI avatar that takes on a role — examiner, hiring manager, patient — and responds in real time: it listens, follows up, and then gives you structured feedback.

How does Trainexus work?

You choose a scenario and hold the conversation by microphone, right in the browser, with a speaking AI avatar. Every word is transcribed live; afterwards you receive structured feedback — inline on individual statements and as an overall assessment, exportable as HTML and PDF. If you also want to analyse presence, voice and body language, use Video Analysis — an optional add-on module.

Who is Trainexus for?

Anyone with an important conversation ahead: students facing exams, job applicants, medical and nursing staff, lawyers, language learners, managers — and institutions that want to scale the training.

Is Trainexus GDPR-compliant?

Yes. Hosting in Germany, data processing agreements with every provider, active deletion after each session, and no use of your data for AI training. Video Analysis runs on local models on a GPU server in Germany — no video leaves Germany, and the raw footage is deleted immediately after processing. Details in the privacy policy.

Can I have my own scenario created?

Yes — in two ways: we set it up for you, or you build it yourself with PromptStudio: describe it in plain language, upload documents, test, publish. No technical knowledge at all.

What does it cost?

From €30 per hour of conversation per person; the premium tier applies to more capable models and special requirements. Work out your own case directly — in the pricing calculator. For an individual quote, book a meeting with us.

Pricing

What does Trainexus cost?

Transparent and fair: from €30 per hour of conversation per person. The premium tier applies to more capable models and special requirements. Work out your scenario directly — in 30-minute steps.

1 person5 people
30 min10 h

Request trial access

You can try up to two scenarios. Pick them in the scenario catalogue and/or describe your own — the chat in the bottom right works too. We will set up your free 10-minute trial access.

Describe your own scenario (optional)
No scenario chosen? Browse the catalogue → or describe your own above.
Or write to us directly: [email protected] · Meeting: termin.trainexus.de