Quickstart

pip install zeroruntime

npm install @zeroruntime/js-sdk

import zeroruntime
from zeroruntime import Agent, Pipeline, EOUConfig, InterruptConfig
from zeroruntime.plugins import DeepgramSTT, GoogleLLM, CartesiaTTS, SileroVAD, RNNoise
from zeroruntime.inference import TurnDetector

AGENT_ID = "assistant"

pipeline = Pipeline(
    stt=DeepgramSTT(),
    llm=GoogleLLM(model="gemini-2.5-flash", max_output_tokens=8192),
    tts=CartesiaTTS(),
    vad=SileroVAD(threshold=0.4),
    turn_detector=TurnDetector(model="echo-large"),
    denoise=RNNoise(),
    eou_config=EOUConfig(mode="ADAPTIVE", min_max_speech_wait_timeout=[0.1, 0.3]),
    interrupt_config=InterruptConfig(
        interrupt_min_duration=0.5,
        interrupt_min_words=2,
        resume_on_false_interrupt=True,
    ),
)


class Assistant(Agent):
    def __init__(self):
        super().__init__(
            agent_id=AGENT_ID,
            instructions="You are a friendly voice assistant. Keep replies short.",
            pipeline=pipeline,
        )

    async def on_enter(self):
        await self.session.say("Hi! How can I help?")

    async def on_exit(self):
        pass


if __name__ == "__main__":
    # Pass the class itself (not an instance): serve() builds a fresh Assistant +
    # pipeline per call, which is required for correct per-call state under concurrent calls.
    zeroruntime.serve(Assistant, on_ready=lambda: zeroruntime.invoke(AGENT_ID, room=zeroruntime.Room(playground=True)))

import * as zeroruntime from '@zeroruntime/js-sdk';
import { Agent, EOUConfig, InterruptConfig, Pipeline, Room } from '@zeroruntime/js-sdk';
import { DeepgramSTT, GoogleLLM, CartesiaTTS, SileroVAD, RNNoise } from '@zeroruntime/js-sdk/plugins';
import { TurnDetector } from '@zeroruntime/js-sdk/inference';

const AGENT_ID = 'assistant';

const pipeline = Pipeline({
  stt: DeepgramSTT(),
  llm: GoogleLLM({ model: 'gemini-2.5-flash', max_output_tokens: 8192 }),
  tts: CartesiaTTS(),
  vad: SileroVAD({ threshold: 0.4 }),
  turn_detector: TurnDetector({ model: 'echo-large' }),
  denoise: RNNoise(),
  eou_config: EOUConfig({ mode: 'ADAPTIVE', min_max_speech_wait_timeout: [0.1, 0.3] }),
  interrupt_config: InterruptConfig({
    interrupt_min_duration: 0.5,
    interrupt_min_words: 2,
    resume_on_false_interrupt: true,
  }),
});


class Assistant extends Agent {
  constructor() {
    super({
      agent_id: AGENT_ID,
      instructions: 'You are a friendly voice assistant. Keep replies short.',
      pipeline,
    });
  }

  async on_enter() {
    await this.session!.say('Hi! How can I help?');
  }
}

// Pass the class itself (not an instance): serve() builds a fresh Assistant +
// pipeline per call, which is required for correct per-call state under concurrent calls.
await zeroruntime.serve(Assistant, {
  on_ready: () => zeroruntime.invoke(AGENT_ID, { room: Room({ playground: true }) }),
});

python agent.py

npx tsx agent.ts     # or compile with tsc and run node agent.js