diff --git a/ansible_ai_connect/ai/api/model_pipelines/http/pipelines.py b/ansible_ai_connect/ai/api/model_pipelines/http/pipelines.py
index 3ce05be77..41b021eb7 100644
--- a/ansible_ai_connect/ai/api/model_pipelines/http/pipelines.py
+++ b/ansible_ai_connect/ai/api/model_pipelines/http/pipelines.py
@@ -16,7 +16,9 @@
 import logging
 from typing import Optional
 
+import aiohttp
 import requests
+from django.http import StreamingHttpResponse
 from health_check.exceptions import ServiceUnavailable
 
 from ansible_ai_connect.ai.api.exceptions import (
@@ -39,6 +41,8 @@
     MetaData,
     ModelPipelineChatBot,
     ModelPipelineCompletions,
+    ModelPipelineStreamingChatBot,
+    StreamingChatBotParameters,
 )
 from ansible_ai_connect.ai.api.model_pipelines.registry import Register
 from ansible_ai_connect.healthcheck.backends import (
@@ -120,13 +124,12 @@ def infer_from_parameters(self, api_key, model_id, context, prompt, suggestion_i
         raise NotImplementedError
 
 
-@Register(api_type="http")
-class HttpChatBotPipeline(HttpMetaData, ModelPipelineChatBot[HttpConfiguration]):
+class HttpChatBotMetaData(HttpMetaData):
 
     def __init__(self, config: HttpConfiguration):
         super().__init__(config=config)
 
-    def invoke(self, params: ChatBotParameters) -> ChatBotResponse:
+    def prepare_data(self, params: ChatBotParameters):
         query = params.query
         conversation_id = params.conversation_id
         provider = params.provider
@@ -142,11 +145,49 @@ def invoke(self, params: ChatBotParameters) -> ChatBotResponse:
             data["conversation_id"] = str(conversation_id)
         if system_prompt:
             data["system_prompt"] = str(system_prompt)
+        return data
+
+    def self_test(self) -> Optional[HealthCheckSummary]:
+        summary: HealthCheckSummary = HealthCheckSummary(
+            {
+                MODEL_MESH_HEALTH_CHECK_PROVIDER: "http",
+                MODEL_MESH_HEALTH_CHECK_MODELS: "ok",
+            }
+        )
+        try:
+            headers = {"Content-Type": "application/json"}
+            r = requests.get(self.config.inference_url + "/readiness", headers=headers)
+            r.raise_for_status()
+
+            data = r.json()
+            ready = data.get("ready")
+            if not ready:
+                reason = data.get("reason")
+                summary.add_exception(
+                    MODEL_MESH_HEALTH_CHECK_MODELS,
+                    HealthCheckSummaryException(ServiceUnavailable(reason)),
+                )
+
+        except Exception as e:
+            logger.exception(str(e))
+            summary.add_exception(
+                MODEL_MESH_HEALTH_CHECK_MODELS,
+                HealthCheckSummaryException(ServiceUnavailable(ERROR_MESSAGE), e),
+            )
+        return summary
+
+
+@Register(api_type="http")
+class HttpChatBotPipeline(HttpChatBotMetaData, ModelPipelineChatBot[HttpConfiguration]):
+
+    def __init__(self, config: HttpConfiguration):
+        super().__init__(config=config)
 
+    def invoke(self, params: ChatBotParameters) -> ChatBotResponse:
         response = requests.post(
             self.config.inference_url + "/v1/query",
             headers=self.headers,
-            json=data,
+            json=self.prepare_data(params),
             timeout=self.timeout(1),
             verify=self.config.verify_ssl,
         )
@@ -171,31 +212,44 @@ def invoke(self, params: ChatBotParameters) -> ChatBotResponse:
             detail = json.loads(response.text).get("detail", "")
             raise ChatbotInternalServerException(detail=detail)
 
-    def self_test(self) -> Optional[HealthCheckSummary]:
-        summary: HealthCheckSummary = HealthCheckSummary(
-            {
-                MODEL_MESH_HEALTH_CHECK_PROVIDER: "http",
-                MODEL_MESH_HEALTH_CHECK_MODELS: "ok",
-            }
-        )
-        try:
-            headers = {"Content-Type": "application/json"}
-            r = requests.get(self.config.inference_url + "/readiness", headers=headers)
-            r.raise_for_status()
 
-            data = r.json()
-            ready = data.get("ready")
-            if not ready:
-                reason = data.get("reason")
-                summary.add_exception(
-                    MODEL_MESH_HEALTH_CHECK_MODELS,
-                    HealthCheckSummaryException(ServiceUnavailable(reason)),
-                )
+class HttpStreamingChatBotMetaData(HttpChatBotMetaData):
 
-        except Exception as e:
-            logger.exception(str(e))
-            summary.add_exception(
-                MODEL_MESH_HEALTH_CHECK_MODELS,
-                HealthCheckSummaryException(ServiceUnavailable(ERROR_MESSAGE), e),
-            )
-        return summary
+    def __init__(self, config: HttpConfiguration):
+        super().__init__(config=config)
+
+    def prepare_data(self, params: StreamingChatBotParameters):
+        data = super().prepare_data(params)
+
+        media_type = params.media_type
+        if media_type:
+            data["media_type"] = str(media_type)
+
+        return data
+
+
+@Register(api_type="http")
+class HttpStreamingChatBotPipeline(
+    HttpStreamingChatBotMetaData, ModelPipelineStreamingChatBot[HttpConfiguration]
+):
+
+    def __init__(self, config: HttpConfiguration):
+        super().__init__(config=config)
+
+    def invoke(self, params: StreamingChatBotParameters) -> StreamingHttpResponse:
+        raise NotImplementedError
+
+    async def async_invoke(self, params: StreamingChatBotParameters) -> StreamingHttpResponse:
+        async with aiohttp.ClientSession(raise_for_status=True) as session:
+            headers = {
+                "Content-Type": "application/json",
+                "Accept": "application/json,text/event-stream",
+            }
+            async with session.post(
+                self.config.inference_url + "/v1/streaming_query",
+                json=self.prepare_data(params),
+                headers=headers,
+            ) as r:
+                async for chunk in r.content:
+                    logger.debug(chunk)
+                    yield chunk
diff --git a/ansible_ai_connect/ai/api/model_pipelines/pipelines.py b/ansible_ai_connect/ai/api/model_pipelines/pipelines.py
index f35f8f79c..8c66f1d53 100644
--- a/ansible_ai_connect/ai/api/model_pipelines/pipelines.py
+++ b/ansible_ai_connect/ai/api/model_pipelines/pipelines.py
@@ -242,6 +242,33 @@ def init(
 ChatBotResponse = Any
 
 
+@define
+class StreamingChatBotParameters(ChatBotParameters):
+    media_type: str
+
+    @classmethod
+    def init(
+        cls,
+        query: str,
+        provider: Optional[str] = None,
+        model_id: Optional[str] = None,
+        conversation_id: Optional[str] = None,
+        system_prompt: Optional[str] = None,
+        media_type: Optional[str] = None,
+    ):
+        return cls(
+            query=query,
+            provider=provider,
+            model_id=model_id,
+            conversation_id=conversation_id,
+            system_prompt=system_prompt,
+            media_type=media_type,
+        )
+
+
+StreamingChatBotResponse = Any
+
+
 class MetaData(Generic[PIPELINE_CONFIGURATION], metaclass=ABCMeta):
 
     def __init__(self, config: PIPELINE_CONFIGURATION):
@@ -274,6 +301,9 @@ def alias() -> str:
     def invoke(self, params: PIPELINE_PARAMETERS) -> PIPELINE_RETURN:
         raise NotImplementedError
 
+    async def async_invoke(self, params: PIPELINE_PARAMETERS) -> PIPELINE_RETURN:
+        raise NotImplementedError
+
     @abstractmethod
     def self_test(self) -> Optional[HealthCheckSummary]:
         raise NotImplementedError
@@ -381,3 +411,17 @@ def __init__(self, config: PIPELINE_CONFIGURATION):
     @staticmethod
     def alias():
         return "chatbot-service"
+
+
+class ModelPipelineStreamingChatBot(
+    ModelPipeline[PIPELINE_CONFIGURATION, ChatBotParameters, StreamingChatBotResponse],
+    Generic[PIPELINE_CONFIGURATION],
+    metaclass=ABCMeta,
+):
+
+    def __init__(self, config: PIPELINE_CONFIGURATION):
+        super().__init__(config=config)
+
+    @staticmethod
+    def alias():
+        return "streaming-chatbot-service"
diff --git a/ansible_ai_connect/ai/api/model_pipelines/registry.py b/ansible_ai_connect/ai/api/model_pipelines/registry.py
index f9ee4c88b..1426099ec 100644
--- a/ansible_ai_connect/ai/api/model_pipelines/registry.py
+++ b/ansible_ai_connect/ai/api/model_pipelines/registry.py
@@ -30,6 +30,7 @@
     ModelPipelinePlaybookGeneration,
     ModelPipelineRoleExplanation,
     ModelPipelineRoleGeneration,
+    ModelPipelineStreamingChatBot,
 )
 from ansible_ai_connect.main.settings.types import t_model_mesh_api_type
 
@@ -45,6 +46,7 @@
         ModelPipelinePlaybookExplanation,
         ModelPipelineRoleExplanation,
         ModelPipelineChatBot,
+        ModelPipelineStreamingChatBot,
         PipelineConfiguration,
         Serializer,
     ]
diff --git a/ansible_ai_connect/ai/api/serializers.py b/ansible_ai_connect/ai/api/serializers.py
index 329e4386b..f3bc7d5a0 100644
--- a/ansible_ai_connect/ai/api/serializers.py
+++ b/ansible_ai_connect/ai/api/serializers.py
@@ -352,6 +352,14 @@ class ChatRequestSerializer(serializers.Serializer):
     )
 
 
+class StreamingChatRequestSerializer(ChatRequestSerializer):
+    media_type = serializers.CharField(
+        required=False,
+        label="Media type",
+        help_text=("A media type to be used in the output from LLM."),
+    )
+
+
 class ReferencedDocumentsSerializer(serializers.Serializer):
     docs_url = serializers.CharField()
     title = serializers.CharField()
diff --git a/ansible_ai_connect/ai/api/versions/v1/ai/urls.py b/ansible_ai_connect/ai/api/versions/v1/ai/urls.py
index 2921b3032..aad4bcdb8 100644
--- a/ansible_ai_connect/ai/api/versions/v1/ai/urls.py
+++ b/ansible_ai_connect/ai/api/versions/v1/ai/urls.py
@@ -25,4 +25,5 @@
     path("generations/role/", views.GenerationRole.as_view(), name="generations/role"),
     path("feedback/", views.Feedback.as_view(), name="feedback"),
     path("chat/", views.Chat.as_view(), name="chat"),
+    path("streaming_chat/", views.StreamingChat.as_view(), name="streaming_chat"),
 ]
diff --git a/ansible_ai_connect/ai/api/versions/v1/ai/views.py b/ansible_ai_connect/ai/api/versions/v1/ai/views.py
index 1667003af..f47679e96 100644
--- a/ansible_ai_connect/ai/api/versions/v1/ai/views.py
+++ b/ansible_ai_connect/ai/api/versions/v1/ai/views.py
@@ -21,6 +21,7 @@
     Feedback,
     GenerationPlaybook,
     GenerationRole,
+    StreamingChat,
 )
 
 __all__ = [
@@ -32,4 +33,5 @@
     "ExplanationRole",
     "Feedback",
     "Chat",
+    "StreamingChat",
 ]
diff --git a/ansible_ai_connect/ai/api/views.py b/ansible_ai_connect/ai/api/views.py
index 02bcc18e2..98716e006 100644
--- a/ansible_ai_connect/ai/api/views.py
+++ b/ansible_ai_connect/ai/api/views.py
@@ -21,6 +21,7 @@
 from attr import asdict
 from django.apps import apps
 from django.conf import settings
+from django.http import StreamingHttpResponse
 from django_prometheus.conf import NAMESPACE
 from drf_spectacular.utils import OpenApiResponse, extend_schema
 from oauth2_provider.contrib.rest_framework import IsAuthenticatedOrTokenHasScope
@@ -77,10 +78,12 @@
     ModelPipelinePlaybookGeneration,
     ModelPipelineRoleExplanation,
     ModelPipelineRoleGeneration,
+    ModelPipelineStreamingChatBot,
     PlaybookExplanationParameters,
     PlaybookGenerationParameters,
     RoleExplanationParameters,
     RoleGenerationParameters,
+    StreamingChatBotParameters,
 )
 from ansible_ai_connect.ai.api.pipelines.completions import CompletionsPipeline
 from ansible_ai_connect.ai.api.telemetry import schema1
@@ -134,6 +137,7 @@
     PlaybookGenerationAction,
     RoleGenerationAction,
     SentimentFeedback,
+    StreamingChatRequestSerializer,
     SuggestionQualityFeedback,
 )
 from .telemetry.schema1 import (
@@ -1126,3 +1130,82 @@ def post(self, request) -> Response:
             status=rest_framework_status.HTTP_200_OK,
             headers=headers,
         )
+
+
+class StreamingChat(AACSAPIView):
+    """
+    Send a message to the backend chatbot service and get a streaming reply.
+    """
+
+    class StreamingChatEndpointThrottle(EndpointRateThrottle):
+        scope = "chat"
+
+    permission_classes = [
+        permissions.IsAuthenticated,
+        IsAuthenticatedOrTokenHasScope,
+        IsRHInternalUser | IsTestUser,
+    ]
+    required_scopes = ["read", "write"]
+    schema1_event = schema1.ChatBotOperationalEvent  # TODO
+    request_serializer_class = StreamingChatRequestSerializer
+    throttle_classes = [StreamingChatEndpointThrottle]
+
+    llm: ModelPipelineStreamingChatBot
+
+    def __init__(self):
+        super().__init__()
+        self.llm = apps.get_app_config("ai").get_model_pipeline(ModelPipelineStreamingChatBot)
+
+        self.chatbot_enabled = (
+            self.llm.config.inference_url
+            and self.llm.config.model_id
+            and settings.CHATBOT_DEFAULT_PROVIDER
+        )
+        if self.chatbot_enabled:
+            logger.debug("Chatbot is enabled.")
+        else:
+            logger.debug("Chatbot is not enabled.")
+
+    @extend_schema(
+        request=StreamingChatRequestSerializer,
+        responses={
+            200: ChatResponseSerializer,  # TODO
+            400: OpenApiResponse(description="Bad request"),
+            403: OpenApiResponse(description="Forbidden"),
+            413: OpenApiResponse(description="Prompt too long"),
+            422: OpenApiResponse(description="Validation failed"),
+            500: OpenApiResponse(description="Internal server error"),
+            503: OpenApiResponse(description="Service unavailable"),
+        },
+        summary="Streaming chat request",
+    )
+    def post(self, request) -> Response:
+        if not self.chatbot_enabled:
+            raise ChatbotNotEnabledException()
+
+        req_query = self.validated_data["query"]
+        req_system_prompt = self.validated_data.get("system_prompt")
+        req_provider = self.validated_data.get("provider", settings.CHATBOT_DEFAULT_PROVIDER)
+        conversation_id = self.validated_data.get("conversation_id")
+        media_type = self.validated_data.get("media_type")
+
+        # Initialise Segment Event early, in case of exceptions
+        self.event.chat_prompt = anonymize_struct(req_query)
+        self.event.chat_system_prompt = req_system_prompt
+        self.event.provider_id = req_provider
+        self.event.conversation_id = conversation_id
+        self.event.modelName = self.req_model_id or self.llm.config.model_id
+
+        return StreamingHttpResponse(
+            self.llm.async_invoke(
+                StreamingChatBotParameters.init(
+                    query=req_query,
+                    system_prompt=req_system_prompt,
+                    model_id=self.req_model_id or self.llm.config.model_id,
+                    provider=req_provider,
+                    conversation_id=conversation_id,
+                    media_type=media_type,
+                )
+            ),
+            content_type="text/event-stream",
+        )
diff --git a/ansible_ai_connect/main/settings/base.py b/ansible_ai_connect/main/settings/base.py
index fb2554d60..4b9e3ac86 100644
--- a/ansible_ai_connect/main/settings/base.py
+++ b/ansible_ai_connect/main/settings/base.py
@@ -55,6 +55,7 @@
 # Application definition
 
 INSTALLED_APPS = [
+    "daphne",
     "django.contrib.admin",
     "django.contrib.auth",
     "django.contrib.contenttypes",
@@ -93,6 +94,11 @@
     "csp.middleware.CSPMiddleware",
 ]
 
+if os.environ.get("CSRF_TRUSTED_ORIGINS"):
+    CSRF_TRUSTED_ORIGINS = os.environ.get("CSRF_TRUSTED_ORIGINS").split(",")
+else:
+    CSRF_TRUSTED_ORIGINS = ["http://localhost:8000"]
+
 # Allow Prometheus to scrape metrics
 ALLOWED_CIDR_NETS = [os.environ.get("ALLOWED_CIDR_NETS", "10.0.0.0/8")]
 
@@ -294,7 +300,7 @@ def is_ssl_enabled(value: str) -> bool:
         },
     },
     "handlers": {
-        "console": {"class": "logging.StreamHandler", "formatter": "simple", "level": "INFO"},
+        "console": {"class": "logging.StreamHandler", "formatter": "simple", "level": "DEBUG"},
     },
     "loggers": {
         "django": {
@@ -334,6 +340,11 @@ def is_ssl_enabled(value: str) -> bool:
             "level": "INFO",
             "propagate": False,
         },
+        "ansible_ai_connect.ai.api.streaming_chat": {
+            "handlers": ["console"],
+            "level": "DEBUG",
+            "propagate": False,
+        },
     },
     "root": {
         "handlers": ["console"],
@@ -358,6 +369,7 @@ def is_ssl_enabled(value: str) -> bool:
 ]
 
 WSGI_APPLICATION = "ansible_ai_connect.main.wsgi.application"
+ASGI_APPLICATION = "ansible_ai_connect.main.asgi.application"
 
 # Database
 # https://docs.djangoproject.com/en/4.1/ref/settings/#databases
diff --git a/ansible_ai_connect/main/settings/legacy.py b/ansible_ai_connect/main/settings/legacy.py
index 6543f43c8..cc7aef87c 100644
--- a/ansible_ai_connect/main/settings/legacy.py
+++ b/ansible_ai_connect/main/settings/legacy.py
@@ -192,6 +192,15 @@ def load_from_env_vars():
             "stream": False,
         },
     }
+    model_pipelines_config["ModelPipelineStreamingChatBot"] = {
+        "provider": "http",
+        "config": {
+            "inference_url": chatbot_service_url or "http://localhost:8000",
+            "model_id": chatbot_service_model_id or "granite3-8b",
+            "verify_ssl": model_service_verify_ssl,
+            "stream": False,
+        },
+    }
 
     # Enable Health Checks where we have them implemented
     model_pipelines_config["ModelPipelineCompletions"]["config"][
diff --git a/ansible_ai_connect/main/views.py b/ansible_ai_connect/main/views.py
index 60ca899bb..d81b6607b 100644
--- a/ansible_ai_connect/main/views.py
+++ b/ansible_ai_connect/main/views.py
@@ -28,7 +28,9 @@
 from rest_framework.renderers import BaseRenderer
 from rest_framework.views import APIView
 
-from ansible_ai_connect.ai.api.model_pipelines.pipelines import ModelPipelineChatBot
+from ansible_ai_connect.ai.api.model_pipelines.pipelines import (
+    ModelPipelineStreamingChatBot,
+)
 from ansible_ai_connect.ai.api.permissions import (
     IsOrganisationAdministrator,
     IsOrganisationLightspeedSubscriber,
@@ -121,12 +123,12 @@ class ChatbotView(ProtectedTemplateView):
         IsRHInternalUser | IsTestUser,
     ]
 
-    llm: ModelPipelineChatBot
+    llm: ModelPipelineStreamingChatBot
     chatbot_enabled: bool
 
     def __init__(self):
         super().__init__()
-        self.llm = apps.get_app_config("ai").get_model_pipeline(ModelPipelineChatBot)
+        self.llm = apps.get_app_config("ai").get_model_pipeline(ModelPipelineStreamingChatBot)
         self.chatbot_enabled = (
             self.llm.config.inference_url
             and self.llm.config.model_id
diff --git a/requirements-aarch64.txt b/requirements-aarch64.txt
index 588da2792..440177130 100644
--- a/requirements-aarch64.txt
+++ b/requirements-aarch64.txt
@@ -7,7 +7,9 @@
 aiohappyeyeballs==2.3.5
     # via aiohttp
 aiohttp==3.10.11
-    # via langchain
+    # via
+    #   -r requirements.in
+    #   langchain
 aiosignal==1.3.1
     # via aiohttp
 annotated-types==0.6.0
@@ -29,7 +31,9 @@ anyio==4.6.2.post1
 argparse==1.4.0
     # via uwsgi-readiness-check
 asgiref==3.8.1
-    # via django
+    # via
+    #   daphne
+    #   django
 asttokens==2.4.1
     # via stack-data
 attrs==23.2.0
@@ -37,6 +41,12 @@ attrs==23.2.0
     #   aiohttp
     #   jsonschema
     #   referencing
+    #   service-identity
+    #   twisted
+autobahn==24.4.2
+    # via daphne
+automat==24.8.1
+    # via twisted
 backcall==0.2.0
     # via ipython
 backoff==2.2.1
@@ -68,13 +78,19 @@ charset-normalizer==3.3.2
     # via requests
 click==8.1.7
     # via black
+constantly==23.10.4
+    # via twisted
 cryptography==43.0.1
     # via
     #   -r requirements.in
     #   ansible-core
+    #   autobahn
     #   jwcrypto
     #   pyopenssl
+    #   service-identity
     #   social-auth-core
+daphne==4.1.2
+    # via -r requirements.in
 decorator==5.1.1
     # via ipython
 defusedxml==0.8.0rc2
@@ -167,13 +183,21 @@ httpx==0.27.2
     # via
     #   langsmith
     #   ollama
+hyperlink==21.0.0
+    # via
+    #   autobahn
+    #   twisted
 idna==3.7
     # via
     #   -r requirements.in
     #   anyio
     #   httpx
+    #   hyperlink
     #   requests
+    #   twisted
     #   yarl
+incremental==24.7.2
+    # via twisted
 inflection==0.5.1
     # via drf-spectacular
 ipython==8.10.0
@@ -307,10 +331,12 @@ pyasn1==0.6.0
     #   oauth2client
     #   pyasn1-modules
     #   rsa
+    #   service-identity
 pyasn1-modules==0.4.0
     # via
     #   google-auth
     #   oauth2client
+    #   service-identity
 pycparser==2.21
     # via cffi
 pydantic==2.9.2
@@ -336,6 +362,7 @@ pyopenssl==24.2.1
     # via
     #   -r requirements.in
     #   pydrive2
+    #   twisted
 pyparsing==3.1.2
     # via httplib2
 pyrfc3339==1.1
@@ -410,6 +437,8 @@ segment-analytics-python==2.2.2
     # via -r requirements.in
 semver==3.0.2
     # via launchdarkly-server-sdk
+service-identity==24.2.0
+    # via twisted
 six==1.16.0
     # via
     #   asttokens
@@ -455,6 +484,10 @@ traitlets==5.14.3
     # via
     #   ipython
     #   matplotlib-inline
+twisted[tls]==24.11.0
+    # via daphne
+txaio==23.1.1
+    # via autobahn
 typing-extensions==4.11.0
     # via
     #   django-test-migrations
@@ -464,6 +497,7 @@ typing-extensions==4.11.0
     #   pydantic
     #   pydantic-core
     #   sqlalchemy
+    #   twisted
 uritemplate==4.1.1
     # via
     #   drf-spectacular
@@ -490,3 +524,8 @@ yamllint==1.35.1
     # via ansible-lint
 yarl==1.17.2
     # via aiohttp
+zope-interface==7.2
+    # via twisted
+
+# The following packages are considered to be unsafe in a requirements file:
+# setuptools
diff --git a/requirements-x86_64.txt b/requirements-x86_64.txt
index 4ca612d3d..284dcb588 100644
--- a/requirements-x86_64.txt
+++ b/requirements-x86_64.txt
@@ -7,7 +7,9 @@
 aiohappyeyeballs==2.3.5
     # via aiohttp
 aiohttp==3.10.11
-    # via langchain
+    # via
+    #   -r requirements.in
+    #   langchain
 aiosignal==1.3.1
     # via aiohttp
 annotated-types==0.6.0
@@ -29,7 +31,9 @@ anyio==4.6.2.post1
 argparse==1.4.0
     # via uwsgi-readiness-check
 asgiref==3.8.1
-    # via django
+    # via
+    #   daphne
+    #   django
 asttokens==2.4.1
     # via stack-data
 attrs==23.2.0
@@ -37,6 +41,12 @@ attrs==23.2.0
     #   aiohttp
     #   jsonschema
     #   referencing
+    #   service-identity
+    #   twisted
+autobahn==24.4.2
+    # via daphne
+automat==24.8.1
+    # via twisted
 backcall==0.2.0
     # via ipython
 backoff==2.2.1
@@ -68,13 +78,19 @@ charset-normalizer==3.3.2
     # via requests
 click==8.1.7
     # via black
+constantly==23.10.4
+    # via twisted
 cryptography==43.0.1
     # via
     #   -r requirements.in
     #   ansible-core
+    #   autobahn
     #   jwcrypto
     #   pyopenssl
+    #   service-identity
     #   social-auth-core
+daphne==4.1.2
+    # via -r requirements.in
 decorator==5.1.1
     # via ipython
 defusedxml==0.8.0rc2
@@ -167,13 +183,21 @@ httpx==0.27.2
     # via
     #   langsmith
     #   ollama
+hyperlink==21.0.0
+    # via
+    #   autobahn
+    #   twisted
 idna==3.7
     # via
     #   -r requirements.in
     #   anyio
     #   httpx
+    #   hyperlink
     #   requests
+    #   twisted
     #   yarl
+incremental==24.7.2
+    # via twisted
 inflection==0.5.1
     # via drf-spectacular
 ipython==8.10.0
@@ -307,10 +331,12 @@ pyasn1==0.6.0
     #   oauth2client
     #   pyasn1-modules
     #   rsa
+    #   service-identity
 pyasn1-modules==0.4.0
     # via
     #   google-auth
     #   oauth2client
+    #   service-identity
 pycparser==2.21
     # via cffi
 pydantic==2.9.2
@@ -336,6 +362,7 @@ pyopenssl==24.2.1
     # via
     #   -r requirements.in
     #   pydrive2
+    #   twisted
 pyparsing==3.1.2
     # via httplib2
 pyrfc3339==1.1
@@ -410,6 +437,8 @@ segment-analytics-python==2.2.2
     # via -r requirements.in
 semver==3.0.2
     # via launchdarkly-server-sdk
+service-identity==24.2.0
+    # via twisted
 six==1.16.0
     # via
     #   asttokens
@@ -455,6 +484,10 @@ traitlets==5.14.3
     # via
     #   ipython
     #   matplotlib-inline
+twisted[tls]==24.11.0
+    # via daphne
+txaio==23.1.1
+    # via autobahn
 typing-extensions==4.11.0
     # via
     #   django-test-migrations
@@ -464,6 +497,7 @@ typing-extensions==4.11.0
     #   pydantic
     #   pydantic-core
     #   sqlalchemy
+    #   twisted
 uritemplate==4.1.1
     # via
     #   drf-spectacular
@@ -490,3 +524,8 @@ yamllint==1.35.1
     # via ansible-lint
 yarl==1.17.2
     # via aiohttp
+zope-interface==7.2
+    # via twisted
+
+# The following packages are considered to be unsafe in a requirements file:
+# setuptools
diff --git a/requirements.in b/requirements.in
index 9f0f237a8..4c71fd087 100644
--- a/requirements.in
+++ b/requirements.in
@@ -9,6 +9,7 @@
 # - https://peps.python.org/pep-0631
 # - https://peps.python.org/pep-0508
 # ======================================================================
+aiohttp==3.10.11
 ansible-anonymizer==1.5.0
 ansible-risk-insight==0.2.7
 ansible-lint==24.2.2
@@ -17,6 +18,7 @@ boto3==1.26.84
 black==24.3.0
 certifi@git+https://github.com/ansible/system-certifi@5aa52ab91f9d579bfe52b5acf30ca799f1a563d9
 cryptography==43.0.1
+daphne==4.1.2
 Django==4.2.18
 django-deprecate-fields==0.1.1
 django-extensions==3.2.1
diff --git a/tools/configs/nginx-wisdom.conf b/tools/configs/nginx-wisdom.conf
index 3f8327679..70c2bb3dd 100644
--- a/tools/configs/nginx-wisdom.conf
+++ b/tools/configs/nginx-wisdom.conf
@@ -2,6 +2,11 @@ upstream uwsgi {
     server unix:///var/run/uwsgi/ansible_wisdom.sock;
 }
 
+upstream daphne {
+ 		server unix:///var/run/daphne/ansible_wisdom.sock;
+}
+
+
 server {
     listen 8000 default_server;
     server_name _;
@@ -14,4 +19,12 @@ server {
         uwsgi_pass uwsgi;
         include /etc/nginx/uwsgi_params;
     }
+
+    location /api/v1/ai/streaming_chat/ {
+        proxy_pass          http://daphne;
+        proxy_http_version  1.1;
+        proxy_set_header    Upgrade $http_upgrade;
+        proxy_set_header    Connection "upgrade";
+        proxy_redirect      off;
+    }
 }
diff --git a/tools/configs/supervisord.conf b/tools/configs/supervisord.conf
index 05a897e5d..44e1826ea 100644
--- a/tools/configs/supervisord.conf
+++ b/tools/configs/supervisord.conf
@@ -28,6 +28,29 @@ stdout_logfile_maxbytes = 0
 stderr_logfile = /dev/stderr
 stderr_logfile_maxbytes = 0
 
+[fcgi-program:daphne]
+# TCP socket used by Nginx backend upstream
+socket=tcp://localhost:9000
+
+# When daphne is running in multiple processes, each needs to have a different socket.
+# In such a case, it is recommended to include process # in the name of socket, but
+# then those generated socket names cannot be specified in nginx config file...
+# So use this with numprocs=1 for now.  See https://github.com/django/daphne/issues/287
+# for more details.
+numprocs=1
+command = /var/www/venv/bin/daphne -u /var/run/daphne/ansible_wisdom.sock --fd 0 --access-log - --proxy-headers ansible_ai_connect.main.asgi:application
+
+autostart = true
+autorestart = true
+stopwaitsecs = 1
+stopsignal = KILL
+stopasgroup = true
+killasgroup = true
+stdout_logfile = /dev/stdout
+stdout_logfile_maxbytes = 0
+stderr_logfile = /dev/stderr
+stderr_logfile_maxbytes = 0
+
 ; [program:test]
 ; command = sleep infinity
 
diff --git a/wisdom-service.Containerfile b/wisdom-service.Containerfile
index 9d20acbf9..88cd80c44 100644
--- a/wisdom-service.Containerfile
+++ b/wisdom-service.Containerfile
@@ -50,7 +50,7 @@ RUN /var/www/venv/bin/python3.11 -m pip --no-cache-dir install --no-binary=all c
 
 RUN /var/www/venv/bin/python3.11 -m pip --no-cache-dir install -r/var/www/ansible-ai-connect-service/requirements.txt
 RUN /var/www/venv/bin/python3.11 -m pip --no-cache-dir install -e/var/www/ansible-ai-connect-service/
-RUN mkdir /var/run/uwsgi
+RUN mkdir /var/run/uwsgi /var/run/daphne
 
 RUN echo -e "\
 {\n\
@@ -99,6 +99,7 @@ RUN for dir in \
       /var/log/supervisor \
       /var/run/supervisor \
       /var/run/uwsgi \
+      /var/run/daphne \
       /var/www/wisdom \
       /var/log/nginx \
       /etc/ari \