Home › Case Studies › AI case study

AI Case study

InterpreterPro: AI that builds a full practice session for court and medical interpreters

InterpreTrain trains court and medical interpreters. We built InterpreterPro, a web app where an interpreter describes what they are preparing for and AI builds a complete, personal practice session in minutes.

Discuss a similar project See it live: app.interpreterpro.ai

Client
InterpreTrain
Industry
Interpreter training (court and medical)
Product
AI web app
Live
app.interpreterpro.ai
Our services
Web development, AI integration, API integration, UI/UX

The challenge

Court and medical interpreters have to keep their skills sharp between assignments, but good practice material is hard to find. Recordings have to match the subject, the speed and the language pair the interpreter actually works in, so most interpreters make their own, which takes hours they do not have.

InterpreTrain wanted a tool that produced this material on demand, so an interpreter could prepare for Monday's hearing on Sunday evening.

What we built

  • A conversational practice builder: the interpreter answers a few questions (setting, language pair, difficulty, expected vocabulary) and the app builds the session.
  • Each session is a set of linked modules: a lesson plan, a vocabulary list from a glossary of vetted legal and medical terms, narrated practice audio at two speeds, and a model interpretation with a bilingual, timestamped transcript.
  • An interpretation review tool where interpreters load their own recording and check their vocabulary in real time.
  • Practice sessions published as hosted web pages: open on any device, no downloads, one link to share with a class. A personal library to save, share and delete practices.
  • Ten full state written mock exams (135 questions each, 9 sections, the real 135-minute clock, pass mark 80%), each also available as 10 short untimed modules.
  • Federal FCICE written exam practice, a readiness assessment, a searchable glossary across legal, medical, education and immigration, and certified practice labs in five languages beyond Spanish.
  • Subscriptions and payments with Stripe, on a Django REST API with JWT login and PostgreSQL.

How the AI works

  • Anthropic Claude writes the practice script, lesson plan and vocabulary set for each session, from the interpreter's answers and the vetted glossary.
  • Google Cloud Text-to-Speech narrates the practice audio, at two speaking speeds so the same material can be practised at rising difficulty.
  • Google Gemini handles transcription, for the bilingual model interpretation and for checking the interpreter's own recordings.
  • Exam questions are vetted content, not AI-generated, so mock exams stay true to the real blueprint.

Results

  • A personal practice session on any legal or medical topic, in minutes instead of hours of self-made material.
  • The same session can be practised at two speeds and compared line by line against a model interpretation.
  • Practices are shareable links, so a teacher can send one session to a whole class.

Technology

  • Anthropic Claude
  • Google Gemini
  • Google Cloud Text-to-Speech
  • Python
  • Django
  • Django REST Framework
  • JWT authentication
  • PostgreSQL
  • JavaScript
  • HTML
  • CSS
  • Stripe

Screens

Practice builder: a short chat about what the interpreter is preparing for
Practice builder: a short chat about what the interpreter is preparing for
A generated practice pack: lesson plan, vocabulary, practice audio, model interpretation
A generated practice pack: lesson plan, vocabulary, practice audio, model interpretation
Model interpretation with a bilingual, timestamped transcript
Model interpretation with a bilingual, timestamped transcript
My Practices: saved sessions on legal and medical topics, ready to share
My Practices: saved sessions on legal and medical topics, ready to share
Ten full state written mock exams, 135 questions each
Ten full state written mock exams, 135 questions each

Practice scripts shown are fictional examples generated in the app.

Questions

What does InterpreterPro do?

An interpreter answers a few questions and the app builds a complete practice session: lesson plan, vocabulary, narrated audio at two speeds and a model interpretation with a bilingual transcript. It also has full written certification mock exams.

Which AI does InterpreterPro use?

Anthropic Claude writes scripts, lesson plans and vocabulary; Google Cloud Text-to-Speech narrates the audio; Google Gemini handles transcription.

Can you build an AI training app for my field?

Yes. We turn your expertise and vetted content into an app where AI builds personal practice for each learner, while exams and key content stay vetted by your experts.

All case studies Portfolio

Tell us what you want to build

Share a few lines about your idea or project. A senior person from our team replies within 1–2 business days, with honest advice on scope, time and cost.

Prefer email? info@arthonsys.com · +91-9887578854

Chat on WhatsApp