Skip to main content
User input enables users to send prompts and requests to AI agents over Ably channels. The agent subscribes to a channel to receive user messages, processes them, and sends responses back. This pattern uses Ably Pub/Sub for realtime, bi-directional communication between users and agents. User input works alongside token streaming patterns to create complete conversational AI experiences. While token streaming handles agent-to-user output, user input handles user-to-agent prompts.

​
How it works

User input follows a channel-based pattern where both users and agents connect to a shared channel:
  1. The agent subscribes to the channel to listen for user messages.
  2. The user publishes a message containing their prompt.
  3. The agent receives the message, processes it, and generates a response.
  4. The agent publishes the response back to the channel, correlating it to the original input.
This decoupled approach means agents don’t need to manage persistent connections to individual users. Instead, they subscribe to channels and respond to messages as they arrive.

​
Identify the user

Agents need to verify that incoming messages are from legitimate users. Use identified clients or user claims to establish a verified identity or role for the user.

​
Verify by user identity

Use the clientId to identify the user who sent a message. This enables personalized responses, per-user rate limiting, or looking up user-specific preferences from your database. When a user authenticates with Ably, embed their identity in the JWT: The clientId is automatically attached to every message the user publishes, so agents can trust this identity.

​
Verify by role

Use user claims to verify that a message comes from a user rather than another agent sharing the channel. This is useful when the agent needs to distinguish message sources without needing the specific user identity. When a user authenticates with Ably, embed their role in the JWT: The user claim is automatically attached to every message the user publishes, so agents can trust this role information.

​
Publish user input

Users publish messages to the channel to send prompts to the agent. Generate a unique promptId for each message to correlate agent responses back to the original prompt.

​
Subscribe to user input

The agent subscribes to a channel to receive messages from users. When a user publishes a message to the channel, the agent receives it through the subscription callback. The following example demonstrates an agent subscribing to receive user input:

​
Publish agent responses

When the agent sends a response, it includes the promptId from the original input so users know which prompt the response relates to. This is especially important when users send multiple prompts in quick succession or when responses are streamed. Use the extras.headers field to include the promptId in agent responses: The user’s client can then match responses to their original prompts:

​
Stream responses

For longer AI responses, you’ll typically want to stream tokens back to the user rather than waiting for the complete response. The promptId correlation allows users to associate streamed tokens with their original prompt. When streaming tokens using message-per-response or message-per-token patterns, include the promptId in the message extras:

​
Handle multiple concurrent prompts

Users may send multiple prompts before receiving responses, especially during long-running AI operations. The correlation pattern ensures responses are matched to the correct prompts:

Build docs developers (and LLMs) love

Get started for freeTalk to us