Rehearsing a Crisis Call Before the Call Is Real

Crisis-support work requires calm, trauma-informed judgment under pressure, but traditional practice often depends on inconsistent peer role-play or learning during live calls. CallCraft exists to create a safer bridge between training and practice: participants can rehearse difficult conversations, make mistakes, receive feedback, and build confidence before real people are relying on them.

This is an independent Prosocial Coding LLC project, separate from my day job. My years in the domestic violence field shaped the standard I built this tool to meet: a practice tool should be just as careful about what it does to a trainee as it is about what it does to a caller. The question that runs through every design decision is the same one I would apply to any crisis service: could this harm someone? A practice space that overclaims what it can prove, or lets a trainee practice unsafe responses without catching them, would fail that test.

What it does

CallCraft is a practice space for crisis-support conversations. A participant picks a scenario, chooses a difficulty level, and enters a practice session with a simulated caller. The session supports structured voice and text rehearsal, live coaching tips, transcripts, saved history, and post-session analysis.

The scoring engine compares a transcript against the scenario's objectives, canonical skill dimensions, and reference criteria on the server to help trainees identify growth areas. A trainee should not be able to read the rubric they are being measured against, and a scenario prompt is kept server-side to maintain evaluation integrity.

Who it is for

This is built for the people who run training, not for a consumer looking for advice in a hard moment: hotline supervisors, training leads, program managers, and institutional teams that need crisis-support practice they can repeat and rely on. The app supplies the reps and the structure, while human trainers keep the call on what matters.

The safety work

The application enforces strict data boundaries and user permissions so that participants operate in separate, secure environments. System requests are validated, and personal information is redacted from transcripts before they are saved. If a session or validation cannot be verified, the system fails closed to protect user privacy. An output safety layer checks simulated call audio and text, interrupting playback if unsafe content or persona breaks are detected, and scoring path errors return structured exceptions rather than incorrect scores.

What it does not prove

CallCraft is a training simulator, not a clinical tool, and it should not be used for diagnosis, certification, or as a replacement for human supervision. It does not contain real survivor or participant data. While system integrity is checked through automated tests before release, these checks do not prove clinical validation. The tool is for learning, practice, and reflection under professional supervision.

Where it is going

Future priorities include clinical review, tighter scenario-authoring controls, stronger adversarial testing, and SSO support for institutional deployments. The goal is to provide a safe, structured repeating space for crisis-support training, with a clean handoff to human trainers.