---
title: "AI + Live Video - Amazon IVS"
description: "Build AI-powered live video experiences with Amazon IVS. Add virtual hosts, style transfer, auto-captions, and intelligent moderation using Amazon Bedrock or other AI providers."
canonical: https://ivs.rocks/ai/
alternates:
  - locale: en
    url: https://ivs.rocks/ai/
  - locale: fr
    url: https://ivs.rocks/fr/ai/
  - locale: ja
    url: https://ivs.rocks/ja/ai/
  - locale: ko
    url: https://ivs.rocks/ko/ai/
---
# AI + Live Video - Amazon IVS

> Build AI-powered live video experiences with Amazon IVS. Add virtual hosts, style transfer, auto-captions, and intelligent moderation using Amazon Bedrock or other AI providers.

The infrastructure for intelligent video

## Amazon IVS delivers the stream.  
You bring the intelligence.

Combine ultra-low latency video infrastructure with AI models from Amazon Bedrock, AWS AI services, or any third-party provider. Create experiences that respond, adapt, and engage in real time.

AI-powered use cases

## Unlock new possibilities

Read more

See it in action

## Demos & code samples

Explore reference implementations and live examples

FeaturedReal-time

### AI video with Decart.ai

Zero-latency video transformation. Mirage LSD delivers instant style transfer with frame-to-frame temporal coherence, while Lucy Edit Live enables prompt-controlled modifications to any element-outfits, backgrounds, objects. Try it live with Amazon IVS Real-time.

Extremely low-latency generation

Near-instant style adaptation

Temporal coherence

Continuous video stream

[Try it live](https://decart.ivsdemos.com/)

 

Real-timeOpen Source

 

### Virtual participant

Add AI-powered participants to real-time sessions. This reference implementation shows how to integrate agents that can see, hear, and respond-creating interactive assistants or automated hosts.

-   Real-time audio/video synthesis
-   Customizable AI personality
-   IVS Real-time integration

[View on GitHub](https://github.com/aws-samples/sample-amazon-ivs-real-time-virtual-participant)

Real-timeOpen Source

 

### WebGPU captions

Client-side caption generation using WebGPU acceleration. No server round-trip. Near-instant speech-to-text that runs entirely in the browser for accessible, low-cost live captioning.

-   GPU-accelerated in-browser
-   Zero server-side cost
-   Privacy-preserving

[View on GitHub](https://github.com/aws-samples/amazon-ivs-webgpu-captions-demo)

Bring your own models

## Flexible AI integration

Amazon IVS can work with most AI providers

### Amazon Bedrock

Foundation models from Amazon and leading AI companies through a unified API. Amazon Nova, Anthropic Claude, Meta Llama, and more.

-   Fully managed
-   Multiple providers
-   Enterprise security

[Learn more](https://aws.amazon.com/bedrock/)

### AWS AI Services

Purpose-built services: Amazon Rekognition for video analysis, Amazon Transcribe for speech-to-text, Amazon Polly for synthesis.

-   Pre-trained models
-   Production-ready
-   Auto-scaling

[Learn more](https://aws.amazon.com/machine-learning/ai-services/)

### Third-party AI

Integrate specialized models from Decart.ai, DeepAR, OpenAI, or deploy your own custom solutions. Mix and match for unique experiences.

-   Specialized capabilities
-   Custom models
-   Partner ecosystem

[View partners](/partners)

Amazon IVS

Real-time and low-latency video infrastructure

Before publish

Filters, effects, pre-processing

Server-side

Transcription, moderation, analysis

Client-side

WebGPU, local inference, overlays

Integrate AI anywhere in your pipeline - IVS handles the video transport.

Why Amazon IVS

## Ready for AI workflows

The infrastructure foundation for AI-powered video experiences

### Real-time latency

Real-time streams fast enough (<300ms) for natural AI interactions. Process frames, run inference, and respond-all within a fraction of a second.

### Global edge network

AWS infrastructure can deliver AI-enhanced experiences worldwide. Low and real-time latency for participants regardless of location.

### Flexible processing

Run AI workloads where it makes sense-client-side with WebGPU, server-side with Bedrock, or combine both for the most demanding usecases.

## Ready to add intelligence to your streams?

Start with our demos, dive into the documentation, or explore AI models on Amazon Bedrock.

[Amazon IVS docs](https://docs.aws.amazon.com/ivs/) [Amazon Bedrock](https://aws.amazon.com/bedrock/) [All demos](/examples)