AI Transport roadmap
AI Transport is under active development with new releases shipping regularly
Ably updates the roadmap as customers say what they need. The fastest way to influence it is to contact Ably, explaining what you are trying to build and where AI Transport can help you.
Delivered
Available today.
| Feature | Description |
|---|---|
| Resumable token streaming | Tokens stream over a session rather than a single HTTP request, so a dropped connection resumes mid-response with every token delivered exactly once, and a reconnecting client catches up automatically. |
| Connection resilience | Realtime delivery over WebSocket with automatic fallback to HTTP streaming and long-polling, so streams survive proxies, firewalls, and unreliable networks. |
| Multi-device session continuity | The same session opens on any device the user picks up, with the full conversation and current state. |
| Persistent history and replay | Conversations survive disconnects and stay available as history, for resumption, audit, and review. |
| Bidirectional control | Any client can signal the agent through the session, so a user can cancel generation, interrupt, send again while the agent responds, and keep interacting rather than only consuming a one-way stream. |
| Conversation branching | Edit and regenerate messages, with each branch kept in the session tree. |
| Client-side tool calls and approval gates | Call client tools over the session, with human-in-the-loop approval where you need it. |
| Drop-in framework integration | A drop-in transport for the Vercel AI SDK, a native codec for the OpenAI Responses API, plus the session hooks for everything else. |
| Shared live state and presence | Agent presence and shared session state exposed directly through the SDK. |
| Durable execution | Pair durable sessions with a workflow engine such as Temporal or Vercel WDK, so a mid-flight process crash retries the failed step cleanly instead of stranding the turn. |
| Enterprise-ready platform | SOC 2 Type II and HIPAA, on Ably's realtime platform. |
Now
Actively building.
| Feature | Description |
|---|---|
| Human and AI handover | Transfer a conversation between an AI and a human agent and back again using the AI Transport SDK, with live supervision and escalation. |
| Anthropic support | A native Anthropic integration, so more of the ecosystem is a drop-in. |
Next
Coming up, and unlikely to change.
| Feature | Description |
|---|---|
| Increased language support | AI Transport SDK implementation in Python. |
| Ably Chat SDK integration | A clean path from chat use cases into AI Transport. |
Future
Planned and exploring. If you care about one of these features then contact Ably, explaining what you are trying to build and where AI Transport can help you.
| Feature | Description |
|---|---|
| Increased language support | AI Transport SDK implementation in Java, Swift, Kotlin, and .NET. |
| Expanded ecosystem support | Native integrations for Gemini, LangChain and LangGraph, TanStack AI, AG-UI, Vue, and Svelte, so more of the ecosystem is a drop-in. |
| Sub-agent orchestration | Better primitives for coordinating multiple sub-agents in a single run. |
| More durable-execution integrations | Inngest and trigger.dev. |
| Multimodal | Audio and video delivery, and integrations with voice providers. |
| Joined history | Make the join between an application's own stored history and the live channel correct on both sides, so every message appears exactly once, for applications that keep their own conversation store. |