Generate voice and chat simulation scripts, roleplay a live caller or agent, and study matching real calls. Grounded in 16 redacted helpline calls and the HVE taxonomy.
Home
Generate
Practice
Library
Settings
Welcome back
Generate a fresh simulation script, practice a live call as the agent or the caller, and keep every session in your Library. Everything stays on this machine.
Recent
Pick options and press Generate. Each press assembles a fresh, varied script from the building blocks.
Library
Your saved scripts and practice sessions, stored on this machine. Reopen, download, play, or delete any of them.
AI engine (local and private)
Script generation always works with no AI and no internet. Turn on Local AI to make the roleplay fully conversational and unlock AI coaching and script enhancement. The model runs on your own machine through the local server, so nothing leaves your infrastructure and no call data is ever sent anywhere. Setup steps are in README_SETUP.md.
Privacy: the AI model runs locally through your own server and Ollama. The app only ever sends the synthetic roleplay text, never the real call transcripts or any caller information. If the server is off, the app falls back to the offline templated roleplay automatically.
Organizations
About
Charming studio v1. Voice profiles derived from 400 redacted real calls (200 AARP and 200 ITRC). Scenarios classified by the HVE Standard taxonomy. Output avoids em dashes and en dashes and keeps the two organizations strictly separate.