Stream

Add customizable, real-time messaging to apps with infrastructure built to support engaging conversations at scale.

Build customizable, low-latency video and voice calls, audio rooms, and livestreams that scale across platforms.

Build responsive, personalized social and notification feeds that help communities discover content and engage in real time.

Help trust and safety teams detect harmful content across media, apply consistent policies, and resolve flagged items with contextual AI review and configurable actions.

Build and deploy responsive, multimodal voice and video AI agents with flexible model, infrastructure, and integration choices.

Manage Stream projects and data from the terminal, automate API workflows, or let an AI agent build and operate Stream integrations.

Help AI coding agents build and extend Stream-powered apps, find current documentation, and work with Stream data and configuration.

Stream Chat Messaging

Stream Chat provides APIs, client and server SDKs, and prebuilt UI components for adding real-time messaging to apps. Build direct messages, group conversations, public channels, livestream chat, and support flows, with customizable interfaces for web, mobile, and gaming platforms. Features include threaded replies, reactions, polls, attachments, moderation, offline support, and push notifications. Its global edge network is designed for low latency and high availability, with support for millions of concurrent users in a channel.

Category: Communications Platform as a Service

• Build Chat Across Platforms
  Use platform-specific SDKs and prebuilt interface components to add chat to web, mobile, and gaming applications.
• Scale Real-Time Conversations Globally
  Run chat on global infrastructure built for low-latency performance and large concurrent audiences, including high-traffic communities and live events.
• Customize Chat To Match Your App
  Adapt chat interfaces to your application's brand and interaction requirements, or build a custom interface using the client SDKs.
• Create Conversations For Any Use Case
  Provide private discussions, group communication, and public channels, with configurable permissions to control participation and channel activity.
• Keep Conversations Interactive
  Give users ways to reply in context, react, vote in polls, and see message activity such as typing and read status.
• Keep Chat State In Sync
  Synchronize conversation data across devices and receive channel events as activity happens.
• Moderate Messages In Real Time
  Apply moderation policies to chat content, block messages that match configured rules, and send flagged or borderline content for review.
• Keep Chat Available Offline
  Allow users to access locally stored conversations while disconnected, with message and interface state syncing when the connection returns.
• Notify Users And Show Presence
  Alert users to relevant chat activity and display online or last-active status to provide visibility into availability.
• Share Rich Content In Chat
  Let users share files, media, links, and custom content within conversations, with URL previews and image resizing supported.
• Translate Messages And Localize Chat
  Help users communicate across languages with message translation and provide localized interface text for chat controls.
• Extend Messages And Commands
  Adapt message content to an app's use case and let users launch configured actions or workflows directly from chat.
• Find Messages Across Conversations
  Use advanced message search to locate conversation content.

Video & Audio

Stream Video & Audio provides APIs and SDKs for building in-app video calling, voice calling, audio rooms, and livestreaming. Calls run on a global edge network, with adaptive media optimization, call quality insights, and options for recording, transcription, captions, and chat integration. Customizable UI components and SDKs support React, iOS, Android, Flutter, React Native, JavaScript, Unity, and other integrations. Livestreams support WebRTC, HLS, and RTMP, and the platform is designed to scale from private calls to large live events.

Category: Communications Platform as a Service

• Build In-App Video Calls
  Add private or group video calling to an application using Stream’s SDKs and API.
• Stream To Large Audiences
  Deliver livestreams for large audiences with support for low-latency WebRTC interactions and HLS playback.
• Reduce Latency Worldwide
  Route participants to nearby servers to reduce latency and support consistent calling performance across regions.
• Launch Interactive Audio Rooms
  Create audio room experiences where participants can request permission to speak, while hosts prepare backstage before going live.
• Optimize Media Automatically
  Adapt video quality to network conditions and display size to help maintain call quality while reducing unnecessary video delivery.
• Build Across Platforms
  Choose SDKs and UI components for supported frameworks, then customize the interface or build a custom UI using lower-level components.
• Scale Calls And Streams
  Build from small calls to large events using infrastructure designed for high participant and viewer volumes.
• Add Reliable Voice Calls
  Give users in-app voice calls supported by a global edge network and redundant audio tracks.
• Record And Transcribe Calls
  Record calls and use transcripts or live captions to capture spoken content.
• Monitor Call Quality
  Review real-time call statistics to understand connection quality, including latency, jitter, packet loss, bitrate, and resolution.
• Recover Calls Seamlessly
  Help calls continue through network disruptions with automatic reconnects and support for transitions between network types.
• Share Screens In Calls
  Enable participants to share their screens during supported calls.
• Connect Chat And Calls
  Add chat messaging to video calling experiences using Stream’s integrated products.

Activity Feeds

Stream Activity Feeds provides APIs and SDKs for building real-time social and notification feeds. It supports timelines, personalized For You feeds, news feeds, Stories, groups, forums, and notification feeds, with configurable ranking, filtering, and activity aggregation. Developers can add posts, comments, reactions, follows, and other engagement features, and subscribe to live updates from web and mobile clients. The platform is designed for high-volume applications and includes integrations for feed moderation and controls such as visibility settings, keyword filters, and approval workflows.

• Personalize And Rank Feed Content
  Surface relevant content using configurable ranking and personalization, including follow-based, interest-based, location-aware, popular, and trending content.
• Enable Rich Social Interactions
  Give users ways to create and engage with feed content through comments, reactions, sharing, reposting, bookmarking, hashtags, mentions, and file uploads.
• Deliver Updates As They Happen
  Keep feed experiences current with live activity, comment, and reaction updates, and deliver notifications through supported real-time channels.
• Build Multiple Feed Experiences
  Use configurable feed types to create social timelines, discovery feeds, Stories, group and forum experiences, and notification centers.
• Scale Feeds With Global Infrastructure
  Run feed experiences on infrastructure designed for high-volume applications, with materialized feeds, a global edge network, published performance benchmarks, and an enterprise uptime SLA.
• Find Relevant Feed Content
  Use flexible filters and queries to discover activities based on attributes, activity type, and full-text search.
• Connect Feeds Through Follows
  Create follow relationships that bring activities from followed feeds into a user's timeline.
• Track And Organize Notifications
  Build notification centers that group related events, display unseen or unread counts, and distinguish notifications that have been seen from those that have been read.
• Integrate Feeds Across Platforms
  Use client and server SDKs for common web and mobile platforms and backend languages to connect feed experiences to applications.
• Manage Feed Content Safety
  Apply visibility controls, keyword filters, and approval workflows, with Stream AI Moderation integration, to manage content in feed-based communities.
• Group Similar Feed Activities
  Keep feeds organized and easier to scan by dynamically grouping similar activity events.

Stream AI Moderation

Stream AI Moderation is an API and moderation workspace for user-generated content from any app. It detects harmful content across text, images, video, audio, and livestreams, and applies shared policies across content types. Teams can configure rules, review flagged items, and use Nora to assess flags in context and take or recommend policy-based actions.

• Moderate Five Content Types With One Policy
  Apply a consistent moderation policy to text, images, video, audio, and livestreams instead of setting up separate moderation systems for each content type.
• Detect 35+ Harms In 50+ Languages
  Identify a broad range of harmful content across supported languages, with ready-made categories and the option to add custom labels.
• Review Flags With User And Conversation Context
  Give moderation decisions context beyond an individual flagged item. Nora can resolve clear cases or prepare recommendations for moderators to approve or change.
• Define And Test Policies Without Model Retraining
  Create and evaluate policy changes against past content before activation, and update rules without waiting for model retraining.
• Build Rules From Labels And User History
  Configure moderation logic without code, combining detected content, user history, and thresholds to fit specific policies.
• Apply Policy-Based Actions To Violations
  Match responses to configured policies, from flagging content for review to removing it or taking action against a user.
• Review Decisions And Handle Appeals
  Give moderators a workflow to examine flagged content, take or revise actions, and handle appeals of content removals.
• Trace Decisions And Their Reasoning
  Review a record of moderation actions and the reasoning behind them to support transparency and reporting.
• Moderate Live Video And Audio In Real Time
  Check live frames and speech while a stream is running, and apply configured actions such as warning a host or ending a stream.
• Process Text In Under 50 Ms
  Check text with reported latency under 50 ms and support high-volume classification workloads.
• Connect Moderation To Any App Or Stream Product
  Use the API with content from an existing app or connect moderation to Stream Chat, Feeds, and Video.

Vision Agents

Vision Agents is an open-source Python framework and developer platform for building, testing, deploying, scaling, and observing real-time voice and video AI agents. It streams audio and video over WebRTC and supports pluggable models, voice services, and vision-processing pipelines. Developers can connect agents to Stream, compatible WebRTC infrastructure, local devices, or private deployments, and add tools, MCP servers, memory, retrieval, and phone calling.

Category: AI Agent Platforms

• Build Real-Time Voice And Video Agents
  Create agents that receive and respond to live audio and video, enabling real-time multimodal applications.
• Choose And Swap AI Providers
  Select providers for language, speech, and vision tasks, and change components to suit an agent's requirements.
• Process Live Video With Vision Models
  Add video understanding and custom computer-vision processing to real-time agent workflows.
• Deploy Agents On Your Chosen Infrastructure
  Choose cloud, on-premises, local, or compatible WebRTC infrastructure rather than being limited to one transport layer.
• Scale And Monitor Production Agents
  Deploy stateless agents for horizontal scaling and expose metrics to monitor production workloads.
• Connect Agents To Tools And APIs
  Enable agents to use external tools and services during live conversations.
• Give Agents Persistent Knowledge
  Use retrieval and stored context to help agents recall information across sessions and respond using knowledge sources.
• Connect Agents To Phone Calls
  Bring agents into inbound and outbound telephony workflows, including calls using SIP trunking.
• Manage Natural Conversation Turns
  Configure listening and turn-taking behavior to support conversational flows that respond to speaker changes and interruptions.
• Update Agents During Live Calls
  Send context or instructions to a running agent without interrupting the live conversation.

Stream CLI

Stream CLI is a terminal tool for managing Stream apps, configuration, users, and data. It can bootstrap projects, support building and debugging apps, and access around 370 API endpoints across Chat, Video, Feeds, and Moderation. Developers can use it directly or let an AI agent run commands and work with Stream. JSON responses, built-in filtering, and nonzero failure exit codes support scripting and CI checks.

• Manage Stream Apps And Project Configuration
  Work with Stream apps and configuration from the terminal, and link individual project folders to the apps they use.
• Set Up Projects And Provision Credentials
  Initialize a project, connect it to an app, configure environment values, and create user tokens through terminal commands.
• Enable Agents To Build And Manage Stream
  Give coding agents terminal access to Stream workflows and install skills that teach agents how to build with Stream SDKs.
• Query Stream APIs And Filter JSON Responses
  Access a broad set of Stream API endpoints and query product data from the terminal, with JSON output and built-in filtering.
• Build And Debug Stream Applications
  Use terminal commands or an AI agent to support application setup, development, and debugging with Stream.
• Automate Checks In Scripts And CI
  Use structured command output, built-in filtering, and failure exit codes to integrate read-only checks into scripts and continuous integration pipelines.

Agent Skills for Stream

Agent Skills for Stream is a Markdown skill pack that gives AI coding agents current context for building with Stream Chat, Video, Feeds, and Moderation. Its router directs requests to specialized skills for app development, live documentation lookup, Stream CLI data and account operations, and platform-specific integrations. Developers can use natural-language prompts to create applications or add features to existing projects, query Stream data, and manage app configuration. The skills are available through a one-command installation and support Claude Code and agents that read the universal skills location, including Cursor and Codex.

• Build Or Extend Apps From Prompts
  Describe an application or feature in plain English and the agent can use Stream skills to build it, including chat, video, activity feeds, and moderation capabilities.
• Build Across Four Stream Products
  Use one skill pack to guide agent work across Chat, Video, Feeds, and Moderation instead of installing separate product-specific skill sets.
• Route Requests To The Right Skill
  Let the router direct build requests, API questions, data operations, and platform-specific work to the relevant specialist included in the pack.
• Find Current SDK Guidance
  Ask an agent how to use Stream APIs and SDKs and receive answers based on live documentation with citations.
• Query Data And Manage Stream Apps
  Delegate supported Stream data queries and app or configuration tasks to an agent using the built-in CLI, or run CLI commands directly in a terminal.
• Build For Web And Native Platforms
  Use platform-specific skills to guide Stream integrations for web and supported native frameworks, including Swift, Android, React Native, Flutter, Unreal, and Unity.
• Guide Chat And Feeds Migrations
  Use specialized migration skills to help move a Sendbird chat app to Stream or generate a Feeds v2-to-v3 sync mapping based on live app data.
• Install Stream Skills With One Command
  Add the skill pack to a supported coding agent with a single installation command.
• Review Open Instructions Before Use
  Install a pack whose skills are Markdown instructions rather than executable installers. The documented migration script is read-only and requires consent, and the skill files are open source.