Inworld AI Agent Runtime

Inworld AI is a realtime AI model and infrastructure company focused on consumer-scale, interactive applications. Founded in 2021 by a team that had led language-model product work at Google DeepMind and built Dialogflow, it first became known for AI-powered non-player characters in games, working with partners like Xbox, Disney, NVIDIA, and Niantic. It has since broadened into a voice AI and agent infrastructure platform for any real-time application, from companions and tutors to enterprise support agents.

At the center is its realtime voice stack. Inworld's text-to-speech generates expressive, natural speech with low latency, voice cloning from seconds of audio, real-time control over emotion and pace, and inline non-verbal sounds like laughs and sighs, across many languages, with the ability to carry conversational context forward so long interactions stop feeling like disconnected scripts. A companion speech-to-text model adds voice profiling, natural turn detection, word-level timestamps, and speaker labels.

Beyond audio, Inworld provides the orchestration layer that ties an agent together. A Realtime API handles full-duplex speech-to-speech with intelligent turn-taking and mid-session tool calling, while a Router exposes a single interface to hundreds of language models from providers like OpenAI, Anthropic, Google, and Mistral, with fallbacks and dynamic selection and no added markup. The Agent Runtime, a low-latency C++ core, coordinates the model, speech, and tools, including external systems through the Model Context Protocol, and notably is free to use, with customers paying only for the model consumption underneath.

Inworld's pitch is closing the gap between prototype and production. Built-in telemetry and the ability to A/B test models, prompts, and pipelines on live traffic let teams tune for engagement and retention without redeploying, and the runtime is designed to scale from a prototype to millions of concurrent users with minimal code changes. It offers SDKs for environments like Node.js and Unreal Engine, a browser playground, and enterprise options including on-premise deployment, data residency, and zero data retention. Typical uses include companion and social apps, language learning and tutoring, health and wellness coaching, AI characters in games and interactive media, and enterprise voice agents for support, sales, and recruiting.