Skip to content

Quickstart

This guide walks through your first conversation. Complete the installation before starting.


Open Conversation Simulator. The home screen shows green status indicators when the conversation engine and the LLM runtime are ready.

If the engine reports a problem instead, the home screen offers a restart button — see Troubleshooting if it persists.


If no model is loaded the home screen shows a “No model loaded” banner. Click Install model or go to Settings → Models.

The in-app model manager:

  1. Lists curated models from the registry with size, license, and hardware requirements.
  2. Shows the license text and asks you to accept before downloading.
  3. Downloads the model file, verifies its SHA-256 checksum, and loads it.

The recommended starter is Qwen3 4B Instruct Q4_K_M (~2.6 GB, Apache-2.0 licensed). It works on machines with as little as 4 GB of GPU VRAM, or runs on CPU with no GPU at all.

No internet connection is needed after the model is downloaded. See local-models.md for all available models and hardware recommendations.


From the home screen, click Browse scenarios (or the scenario icon in the sidebar).

Some scenarios to get started:

PackScenarioWhat it practices
Job Interview BasicsStandard behavioral interviewSTAR responses under pressure
Everyday NegotiationUsed car negotiationOpening offers, counteroffers
Language CaféSpanish small talkCasual vocabulary, greetings
Difficult ConversationsGiving critical feedbackStaying calm, being specific

Click a scenario card to see its description, then click Start conversation.


The conversation screen shows:

  • The NPC’s current dialogue
  • A text input at the bottom for your responses
  • A turn counter and scenario progress bar

Type your response and press Enter (or click Send). The NPC responds in one to five seconds, depending on your hardware and model.

The scenario ends automatically when the NPC reaches a terminal state. You can also click End conversation at any time.


After the conversation the debrief screen shows:

  • A turn-by-turn transcript
  • State changes tracked during play (e.g., pressure +8, patience −3)
  • Rubric scores where the scenario defines them
  • Suggested follow-up scenarios or remixes

You can export the transcript to a text file from the debrief screen. Transcripts are stored locally in ~/.convsim/db/ and are never uploaded anywhere.


  • Remix the scenario — adjust NPC difficulty or starting state from the scenario setup screen before starting.
  • Create a custom scenario — see the scenario authoring guide.
  • Upgrade your model — if responses feel slow or generic, try a larger model; see local-models.md.

KeyAction
EnterSend message
Shift+EnterNew line in input