Simulation Trainer

Practise the hardest conversations of your job with AI that plays every role and gives feedback.

What it is

Simulation Trainer (working name) is an idea for AI role-play simulations in professional training. It is a modern revival of Simchild, an earlier project that simulated child welfare investigations for social work education. AI plays every role in a scenario (parents, child, colleagues), responds freely instead of following a script, and afterwards gives feedback against an assessment rubric.

Who it is for

The first vertical would be sports coaches and club leaders practising hard conversations: an angry parent at the sideline, a player who wants to quit, a suspicion of bullying. Feedback would build on Swedish sports federation guidelines. The same engine could serve other verticals:

• Social services: child welfare investigations and conversations with parents
• Teachers: parent meetings and student welfare cases
• Healthcare: conversations with relatives and delivering difficult news
• Managers and HR: rehabilitation, dismissal and performance conversations
• Housing cooperative boards: annual meetings and neighbour conflicts

How AI does the work

One engine with swappable scenarios, roles and assessment rubrics, sold as vertical products. It could also run directly in Claude or ChatGPT via MCP.

Status: Parked

In the idea bank. Not built or tested against the market yet.

Part of The Unnamed Roads, an AI-native venture studio. See how the studio starts and tests ideas with AI or all projects.