Skip to main content

Quickstart

This guide takes you from zero to a fully analysed call in about five minutes against a locally running backend. For a production setup see Deployment.

Prerequisites​

  • Java 21 (java -version)
  • Maven 3.9+ (mvn -version)
  • Docker (for the Postgres container) or an existing Postgres instance
  • ffmpeg on your $PATH (required for audio preprocessing)
  • A .m4a / .mp3 / .wav recording to analyse

1. Start dependencies​

docker run -d --name scryon-pg \
  -e POSTGRES_USER=scryon \
  -e POSTGRES_PASSWORD=scryon \
  -e POSTGRES_DB=scryon \
  -p 5432:5432 \
  postgres:16

2. Configure secrets​

Create a .env file (or export these vars) — only LEMONFOX_API_KEY and LLM_API_KEY are mandatory for the basic flow.

export LEMONFOX_API_KEY=sk-...
export LLM_API_KEY=sk-...
# Optional but recommended:
export PYANNOTE_ENABLED=true
export PYANNOTE_API_KEY=...

3. Run the backend​

cd scryon-backend
./mvnw spring-boot:run

The server starts on http://localhost:8080. Confirm with:

curl -s http://localhost:8080/api/health | jq

4. Submit a call​

curl -X POST http://localhost:8080/api/calls/analyze \
  -F "file=@/path/to/recording.m4a" \
  -F 'metadata={
        "title": "Demo call",
        "contactName": "Acme Corp",
        "direction": "OUTGOING"
      };type=application/json'

Response:

{
  "callId": "f0a1d2e3-...",
  "status": "QUEUED"
}

5. Poll for completion​

curl -s "http://localhost:8080/api/calls/status?ids=f0a1d2e3-..." | jq

Status transitions are QUEUED → TRANSCRIBING → ANALYZING → COMPLETED. Typical end-to-end latency for a 5-minute call is 10–40 seconds.

6. Read the output​

Once status is COMPLETED:

# Speaker-attributed transcript
curl -s http://localhost:8080/api/calls/$CALL_ID/transcript | jq

# Structured analysis with action items
curl -s http://localhost:8080/api/calls/$CALL_ID/analysis | jq

# Action items as first-class objects
curl -s http://localhost:8080/api/actions | jq

What's next​