> For a complete index of all SignalWire documentation pages, fetch https://signalwire.com/docs/llms.txt # AIChatClient > Async client for holding a text conversation with a SignalWire AI agent from your own server. [chatgateway]: /docs/server-sdks/reference/python/agents/chat-gateway [chat-endpoint]: /docs/apis/rest/ai-chat/chat-methods [error-codes]: /docs/apis/error-codes [createconversation]: /docs/server-sdks/reference/python/agents/ai-chat-client/create-conversation [chat]: /docs/server-sdks/reference/python/agents/ai-chat-client/chat [end]: /docs/server-sdks/reference/python/agents/ai-chat-client/end [delete]: /docs/server-sdks/reference/python/agents/ai-chat-client/delete [log]: /docs/server-sdks/reference/python/agents/ai-chat-client/log [summarize]: /docs/server-sdks/reference/python/agents/ai-chat-client/summarize [rawpost]: /docs/server-sdks/reference/python/agents/ai-chat-client/raw-post [close]: /docs/server-sdks/reference/python/agents/ai-chat-client/close `AIChatClient` sends messages to an AI agent and returns its replies. The agent is the same one a phone call would reach, and each call to [`chat()`][chat] runs one full turn. It is async-first. A turn waits on a full model round trip, measured in seconds, and the typical consumers run on an asyncio event loop where a blocking HTTP call would stall every other conversation. Use it as an async context manager, or call [`close()`][close] yourself. ```bash pip install signalwire-sdk ``` This class holds your API token, so it belongs on a server you control. To reach an agent from a browser, use [`ChatGateway`][chatgateway] instead. ## **Parameters** **`project`** `str | None` Your project ID. Falls back to `SIGNALWIRE_PROJECT_ID`. Raises `ValueError` when neither is set. --- **`token`** `str | None` An API token with the `chat` scope. Falls back to `SIGNALWIRE_API_TOKEN`. --- **`space`** `str | None` Your Space name, used to build the service URL. Falls back to `SIGNALWIRE_SPACE`. --- **`url`** `str | None` The service URL, used verbatim. Overrides `space`. Raises `ValueError` when neither is available. --- **`session`** `aiohttp.ClientSession | None` An existing session to send requests on. When omitted, the client creates and owns one, and `close()` closes it. Pass your own to control connection pooling or timeouts. --- The client does not add Basic authentication or its User-Agent to a session you supply. Configure those headers on the session before passing `session=`, or let the client create the session. A configured environment needs no arguments at all. Identity travels in HTTP Basic auth, never in the request body. ## **Return types** **`ConversationInfo`** `dataclass` Returned by [`create_conversation()`][createconversation]. Carries `id`, `status`, and `initial_message`. --- **`ChatResponse`** `dataclass` Returned by [`chat()`][chat]. Carries `text`, `conversation_id`, and `user_event`. --- **`ChatLog`** `dataclass` Returned by [`log()`][log]. Carries `messages` and `call_timeline`. --- ## **Exceptions** All inherit from `AIChatError`, which carries `code` and `message`. Codes without a specific class raise `AIChatError` itself. Documented codes are listed under [AI chat errors][error-codes]. **`AuthenticationError`** `AIChatError` A JSON-RPC `-32009` response. A public-endpoint HTTP 401 currently raises `AIChatError` instead. --- **`ConversationNotFoundError`** `AIChatError` No conversation with that id exists in your project. --- **`RateLimitError`** `AIChatError` The service returned a JSON-RPC rate-limit error. --- **`ChatInProgressError`** `AIChatError` The service returned a JSON-RPC error indicating that a chat operation is already in progress. --- **`SummaryError`** `AIChatError` Summary generation failed. `code` is `None`, because this failure rides the success envelope rather than arriving as a JSON-RPC error. --- The client maps JSON-RPC error codes to these classes. Platform errors do not always use a JSON-RPC envelope, so an HTTP failure can raise `AIChatError` directly or, for some JSON response shapes, produce an empty result. Check the API error when handling failed calls. ## **Timeouts** The default session has no client-side total cap, a 10-second connect timeout, and a 60-second `sock_read` timeout. The public endpoint separately limits an individual request to 30 seconds; a slower turn returns the same 502 response as an unavailable service. ## **Methods** #### [create\_conversation](/docs/server-sdks/reference/python/agents/ai-chat-client/create-conversation) Create a conversation, or reset an existing one. #### [chat](/docs/server-sdks/reference/python/agents/ai-chat-client/chat) Send a message and return the agent's reply. #### [end](/docs/server-sdks/reference/python/agents/ai-chat-client/end) End the conversation and start post-processing. #### [delete](/docs/server-sdks/reference/python/agents/ai-chat-client/delete) Remove the conversation and its data. #### [log](/docs/server-sdks/reference/python/agents/ai-chat-client/log) Read the conversation back. #### [summarize](/docs/server-sdks/reference/python/agents/ai-chat-client/summarize) Generate a summary of the conversation. #### [raw\_post](/docs/server-sdks/reference/python/agents/ai-chat-client/raw-post) Access one JSON-RPC HTTP response with its body unread. #### [close](/docs/server-sdks/reference/python/agents/ai-chat-client/close) Close the session the client created. ## **Example** ```python {10,13,18} import asyncio from signalwire.ai_chat import AIChatClient CONFIG_URL = "https://bayview-taxi.example.com/swml" async def main(): async with AIChatClient(space="your-space") as client: info = await client.create_conversation("chat-8f21", config_url=CONFIG_URL) print("Ada:", info.initial_message) reply = await client.chat( "chat-8f21", "How much is a van from 123 Gough Street to the airport?" ) print("Ada:", reply.text) await client.end("chat-8f21") asyncio.run(main()) ``` The [AI chat endpoint][chat-endpoint] documents the wire protocol underneath this class.