Merge pull request #3931 from pipecat-ai/aleix/examples-always-use-user-role

Update foundational examples to use system_instruction

changelog/3931.other.md

@@ -0,0 +1 @@
+- Updated foundational examples and eval scripts to use `"user"` role instead of `"system"` when adding messages to `LLMContext`, since system prompts should be set via `system_instruction` on the LLM service.

examples/foundational/02-llm-say-one-thing.py

@@ -60,7 +60,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     @transport.event_handler("on_client_connected")
     async def on_client_connected(transport, client):
         context = LLMContext()
-        context.add_message({"role": "system", "content": "Say hello to the world."})
+        context.add_message({"role": "user", "content": "Say hello to the world."})
         await task.queue_frames([LLMContextFrame(context), EndFrame()])
 
     runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)

examples/foundational/04-transports-small-webrtc.py

@@ -109,7 +109,7 @@ async def run_example(webrtc_connection: SmallWebRTCConnection):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/04a-transports-daily.py

@@ -93,7 +93,7 @@ async def main():
             await transport.capture_participant_transcription(participant["id"])
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/05-sync-speech-and-image.py

@@ -152,7 +152,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         ]:
             messages = [
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": f"Describe a nature photograph suitable for use in a calendar, for the month of {month}. Include only the image description with no preamble. Limit the description to one sentence, please.",
                 }
             ]

examples/foundational/05a-local-sync-speech-and-image.py

@@ -49,7 +49,7 @@ async def main():
         async def get_month_data(month):
             messages = [
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": f"Describe a nature photograph suitable for use in a calendar, for the month of {month}. Include only the image description with no preamble. Limit the description to one sentence, please.",
                 }
             ]

examples/foundational/06-listen-and-respond.py

@@ -129,7 +129,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07-interruptible-cartesia-http.py

@@ -103,7 +103,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07-interruptible.py

@@ -99,7 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07a-interruptible-speechmatics-vad.py

@@ -149,7 +149,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         async def on_client_connected(transport, client):
             logger.info(f"Client connected")
             # Kick off the conversation.
-            context.add_message({"role": "system", "content": "Say a short hello to the user."})
+            context.add_message({"role": "user", "content": "Say a short hello to the user."})
             await task.queue_frames([LLMRunFrame()])
 
         @transport.event_handler("on_client_disconnected")

examples/foundational/07a-interruptible-speechmatics.py

@@ -129,7 +129,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         async def on_client_connected(transport, client):
             logger.info(f"Client connected")
             # Kick off the conversation.
-            context.add_message({"role": "system", "content": "Say a short hello to the user."})
+            context.add_message({"role": "user", "content": "Say a short hello to the user."})
             await task.queue_frames([LLMRunFrame()])
 
         @transport.event_handler("on_client_disconnected")

examples/foundational/07c-interruptible-deepgram-flux.py

@@ -109,7 +109,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07c-interruptible-deepgram-http.py

@@ -104,7 +104,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07c-interruptible-deepgram-vad.py

@@ -106,7 +106,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07c-interruptible-deepgram.py

@@ -100,7 +100,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07d-interruptible-elevenlabs-http.py

@@ -108,7 +108,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07d-interruptible-elevenlabs.py

@@ -100,7 +100,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07f-interruptible-azure-http.py

@@ -102,7 +102,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07f-interruptible-azure.py

@@ -102,7 +102,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07g-interruptible-openai-http.py

@@ -106,7 +106,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07g-interruptible-openai.py

@@ -108,7 +108,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07h-interruptible-openpipe.py

@@ -103,7 +103,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07i-interruptible-xtts.py

@@ -104,7 +104,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07j-interruptible-gladia-vad.py

@@ -114,7 +114,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07j-interruptible-gladia.py

@@ -109,7 +109,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07k-interruptible-lmnt.py

@@ -99,7 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07l-interruptible-groq.py

@@ -95,7 +95,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07n-interruptible-gemini-image.py

@@ -124,7 +124,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation with a styled introduction
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07n-interruptible-gemini.py

@@ -128,7 +128,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": "You are an AI assistant. You can help with a variety of tasks. Introduce yourself and ask the user what they would like to know.",
             }
         )

examples/foundational/07n-interruptible-google-http.py

@@ -111,7 +111,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07n-interruptible-google.py

@@ -111,7 +111,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07o-interruptible-assemblyai-turn-detection.py

@@ -153,7 +153,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07o-interruptible-assemblyai.py

@@ -102,7 +102,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07p-interruptible-krisp-viva.py

@@ -134,7 +134,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07p-interruptible-krisp.py

@@ -103,7 +103,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07q-interruptible-rime-http.py

@@ -107,7 +107,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07q-interruptible-rime.py

@@ -99,7 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07r-interruptible-nvidia.py

@@ -95,7 +95,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07s-interruptible-google-audio-in.py

@@ -218,11 +218,10 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         api_key=os.getenv("GOOGLE_API_KEY"),
         settings=GoogleLLMSettings(
             model="gemini-2.5-flash",
+            system_instruction=system_message,
+            # force a certain amount of thinking if you want it
+            # thinking=GoogleLLMService.ThinkingConfig(thinking_budget=4096)
         ),
-        # force a certain amount of thinking if you want it
-        # params=GoogleLLMService.InputParams(
-        #     thinking=GoogleLLMService.ThinkingConfig(thinking_budget=4096)
-        # ),
     )
 
     tts = GoogleTTSService(
@@ -234,18 +233,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         credentials=os.getenv("GOOGLE_TEST_CREDENTIALS"),
     )
 
-    messages = [
+    context = LLMContext()
-        {
-            "role": "system",
-            "content": system_message,
-        },
-        {
-            "role": "user",
-            "content": "Start by saying hello.",
-        },
-    ]
-
-    context = LLMContext(messages)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
@@ -281,7 +269,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        messages.append({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07t-interruptible-fish.py

@@ -100,7 +100,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07v-interruptible-neuphonic-http.py

@@ -105,7 +105,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07v-interruptible-neuphonic.py

@@ -99,7 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07w-interruptible-fal.py

@@ -106,7 +106,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07x-interruptible-local.py

@@ -82,7 +82,7 @@ async def main():
         ),
     )
 
-    context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+    context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
     await task.queue_frames([LLMRunFrame()])
 
     runner = PipelineRunner()

examples/foundational/07y-interruptible-minimax.py

@@ -107,7 +107,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07z-interruptible-sarvam-http.py

@@ -109,7 +109,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07z-interruptible-sarvam.py

@@ -103,7 +103,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
         # Optionally, you can wait for 30 seconds and then change the voice.

examples/foundational/07za-interruptible-soniox.py

@@ -105,7 +105,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zb-interruptible-inworld-http.py

@@ -112,7 +112,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info("Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07zb-interruptible-inworld.py

@@ -108,7 +108,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info("Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zc-interruptible-asyncai-http.py

@@ -105,7 +105,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             logger.info(f"Client connected")
             # Kick off the conversation.
             context.add_message(
-                {"role": "system", "content": "Please introduce yourself to the user."}
+                {"role": "user", "content": "Please introduce yourself to the user."}
             )
             await task.queue_frames([LLMRunFrame()])
 

examples/foundational/07zc-interruptible-asyncai.py

@@ -100,7 +100,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zd-interruptible-aicoustics.py

@@ -128,7 +128,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         logger.info(f"Client connected")
         await audiobuffer.start_recording()
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @audiobuffer.event_handler("on_audio_data")

examples/foundational/07ze-interruptible-hume.py

@@ -113,7 +113,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             "💡 Word timestamps are enabled! Watch the console for TTSTextFrame logs showing each word with its PTS."
         )
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zf-interruptible-gradium.py

@@ -106,7 +106,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zg-interruptible-camb.py

@@ -99,7 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     @transport.event_handler("on_client_connected")
     async def on_client_connected(transport, client):
         logger.info("Client connected")
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zi-interruptible-piper.py

@@ -98,7 +98,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zj-interruptible-kokoro.py

@@ -98,7 +98,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/07zk-interruptible-resemble.py

@@ -102,7 +102,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/08-custom-frame-processor.py

@@ -141,7 +141,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected: {client}")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/10-wake-phrase.py

@@ -104,7 +104,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": "Please introduce yourself. Tell the user they should say 'Hey Robot' before talking to you.",
             }
         )

examples/foundational/14a-function-calling-anthropic.py

@@ -138,6 +138,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/14d-function-calling-anthropic-video.py

@@ -164,7 +164,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/14d-function-calling-aws-video.py

@@ -169,7 +169,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/14d-function-calling-gemini-flash-video.py

@@ -164,7 +164,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/14d-function-calling-moondream-video.py

@@ -202,7 +202,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/14d-function-calling-openai-video.py

@@ -163,7 +163,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/14i-function-calling-fireworks.py

@@ -136,7 +136,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/15-switch-voices.py

@@ -172,7 +172,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user and let them know the voices you can do. Your initial responses should be as if you were a {tts.current_voice}.",
             }
         )

examples/foundational/15a-switch-languages.py

@@ -160,7 +160,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation.
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user and let them know the languages you speak. Your initial responses should be in {tts.current_language}.",
             }
         )

examples/foundational/16-gpu-container-local-bot.py

@@ -109,7 +109,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     # Handle "latency-ping" messages. The client will send app messages that look like

examples/foundational/17-detect-user-idle.py

@@ -60,14 +60,14 @@ class IdleHandler:
         if self._retry_count == 1:
             # First attempt: Add a gentle prompt to the conversation
             message = {
-                "role": "system",
+                "role": "user",
                 "content": "The user has been quiet. Politely and briefly ask if they're still there.",
             }
             await aggregator.push_frame(LLMMessagesAppendFrame([message], run_llm=True))
         elif self._retry_count == 2:
             # Second attempt: More direct prompt
             message = {
-                "role": "system",
+                "role": "user",
                 "content": "The user is still inactive. Ask if they'd like to continue our conversation.",
             }
             await aggregator.push_frame(LLMMessagesAppendFrame([message], run_llm=True))
@@ -209,7 +209,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
         await asyncio.sleep(30)
         logger.info(f"Disabling idle detection")

examples/foundational/20a-persistent-context-openai.py

@@ -95,12 +95,7 @@ async def load_conversation(params: FunctionCallParams):
         await params.result_callback({"success": False, "error": str(e)})
 
 
-messages = [
+system_instruction = "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way."
-    {
-        "role": "system",
-        "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
-    },
-]
 
 weather_function = FunctionSchema(
     name="get_current_weather",
@@ -185,7 +180,10 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         ),
     )
 
-    llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        system_instruction=system_instruction,
+    )
 
     # you can either register a single function for all function calls, or specific functions
     # llm.register_function(None, fetch_weather_from_api)
@@ -194,7 +192,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     llm.register_function("get_saved_conversation_filenames", get_saved_conversation_filenames)
     llm.register_function("load_conversation", load_conversation)
 
-    context = LLMContext(messages, tools)
+    context = LLMContext(tools=tools)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),

examples/foundational/20b-persistent-context-openai-realtime-beta.py

@@ -31,7 +31,6 @@ from pipecat.services.openai_realtime_beta import (
     SessionProperties,
     TurnDetection,
 )
-from pipecat.services.openai_realtime_beta.events import AudioConfiguration, AudioInput
 from pipecat.transports.base_transport import BaseTransport, TransportParams
 from pipecat.transports.daily.transport import DailyParams
 from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
@@ -182,16 +181,12 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
 
     session_properties = SessionProperties(
-        audio=AudioConfiguration(
+        input_audio_transcription=InputAudioTranscription(),
-            input=AudioInput(
+        # Set openai TurnDetection parameters. Not setting this at all will turn
-                transcription=InputAudioTranscription(),
+        # it on by default
-                # Set openai TurnDetection parameters. Not setting this at all will turn it
+        turn_detection=TurnDetection(silence_duration_ms=1000),
-                # on by default
+        # Or set to False to disable openai turn detection and use transport VAD
-                turn_detection=TurnDetection(silence_duration_ms=1000),
+        # turn_detection=False,
-                # Or set to False to disable openai turn detection and use transport VAD
-                # turn_detection=False,
-            )
-        ),
         # tools=tools,
         instructions="""Your knowledge cutoff is 2023-10. You are a helpful and friendly AI.
 

examples/foundational/20c-persistent-context-anthropic.py

@@ -26,7 +26,7 @@ from pipecat.processors.aggregators.llm_response_universal import (
 )
 from pipecat.runner.types import RunnerArguments
 from pipecat.runner.utils import create_transport
-from pipecat.services.anthropic.llm import AnthropicLLMService
+from pipecat.services.anthropic.llm import AnthropicLLMService, AnthropicLLMSettings
 from pipecat.services.cartesia.tts import CartesiaTTSService, CartesiaTTSSettings
 from pipecat.services.deepgram.stt import DeepgramSTTService
 from pipecat.services.llm_service import FunctionCallParams
@@ -38,7 +38,6 @@ load_dotenv(override=True)
 
 
 BASE_FILENAME = "/tmp/pipecat_conversation_"
-tts = None
 
 
 async def fetch_weather_from_api(params: FunctionCallParams):
@@ -82,7 +81,6 @@ async def save_conversation(params: FunctionCallParams):
 
 
 async def load_conversation(params: FunctionCallParams):
-    global tts
     filename = params.arguments["filename"]
     logger.debug(f"loading conversation from {filename}")
     try:
@@ -96,18 +94,7 @@ async def load_conversation(params: FunctionCallParams):
         await params.result_callback({"success": False, "error": str(e)})
 
 
-# Test message munging ...
+system_instruction = "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a succinct, creative and helpful way. Prefer responses that are one sentence long unless you are asked for a longer or more detailed response."
-messages = [
-    {
-        "role": "system",
-        "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a succinct, creative and helpful way. Prefer responses that are one sentence long unless you are asked for a longer or more detailed response.",
-    },
-    {"role": "user", "content": "Start the call by saying the word 'hello'. Say only that word."},
-    # {"role": "user", "content": ""},
-    # {"role": "assistant", "content": []},
-    # {"role": "user", "content": "Tell me"},
-    # {"role": "user", "content": "a joke"},
-]
 
 weather_function = FunctionSchema(
     name="get_current_weather",
@@ -183,8 +170,6 @@ transport_params = {
 async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     logger.info(f"Starting bot")
 
-    global tts
-
     stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
 
     tts = CartesiaTTSService(
@@ -194,7 +179,10 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         ),
     )
 
-    llm = AnthropicLLMService(api_key=os.getenv("ANTHROPIC_API_KEY"))
+    llm = AnthropicLLMService(
+        api_key=os.getenv("ANTHROPIC_API_KEY"),
+        settings=AnthropicLLMSettings(system_instruction=system_instruction),
+    )
 
     # you can either register a single function for all function calls, or specific functions
     # llm.register_function(None, fetch_weather_from_api)
@@ -203,7 +191,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     llm.register_function("get_saved_conversation_filenames", get_saved_conversation_filenames)
     llm.register_function("load_conversation", load_conversation)
 
-    context = LLMContext(messages, tools)
+    context = LLMContext(tools=tools)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
@@ -234,6 +222,12 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
+        context.add_message(
+            {
+                "role": "user",
+                "content": "Start the call by saying the word 'hello'. Say only that word.",
+            }
+        )
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/20d-persistent-context-gemini.py

@@ -122,11 +122,7 @@ async def load_conversation(params: FunctionCallParams):
         await params.result_callback({"success": False, "error": str(e)})
 
 
-# Test message munging ...
+system_instruction = """You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your
-messages = [
-    {
-        "role": "system",
-        "content": """You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your
 capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that
 can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative
 and helpful way.
@@ -151,13 +147,7 @@ indicate you should use the get_image tool are:
   - Tell me about what you see.
   - Tell me something interesting about what you see.
   - What's happening in the video?
-        """,
+"""
-    },
-    # {"role": "user", "content": ""},
-    # {"role": "assistant", "content": []},
-    # {"role": "user", "content": "Tell me"},
-    # {"role": "user", "content": "a joke"},
-]
 
 weather_function = FunctionSchema(
     name="get_current_weather",
@@ -262,7 +252,10 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         ),
     )
 
-    llm = GoogleLLMService(api_key=os.getenv("GOOGLE_API_KEY"))
+    llm = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
+        system_instruction=system_instruction,
+    )
 
     # you can either register a single function for all function calls, or specific functions
     # llm.register_function(None, fetch_weather_from_api)
@@ -272,7 +265,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     llm.register_function("load_conversation", load_conversation)
     llm.register_function("get_image", get_image)
 
-    context = LLMContext(messages, tools)
+    context = LLMContext(tools=tools)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
@@ -308,9 +301,9 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         client_id = get_transport_client_id(transport, client)
 
         # Kick off the conversation.
-        messages.append(
+        context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": f"Please introduce yourself to the user. Use '{client_id}' as the user ID during function calls.",
             }
         )

examples/foundational/20e-persistent-context-aws-nova-sonic.py

@@ -235,12 +235,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     llm.register_function("get_saved_conversation_filenames", get_saved_conversation_filenames)
     llm.register_function("load_conversation", load_conversation)
 
-    context = LLMContext(
+    context = LLMContext(tools=tools)
-        messages=[
-            {"role": "user", "content": "Hello!"},
-        ],
-        tools=tools,
-    )
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(context)
 
     pipeline = Pipeline(
@@ -266,6 +261,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
         # HACK: if using the older Nova Sonic (pre-2) model, you need this special way of
         # triggering the first assistant response. Note that this trigger requires a special

examples/foundational/21-tavus-transport.py

@@ -97,7 +97,7 @@ async def main():
             # Kick off the conversation.
             context.add_message(
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": "Start by greeting the user and ask how you can help.",
                 }
             )

examples/foundational/21a-tavus-video-service.py

@@ -115,7 +115,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             # Kick off the conversation.
             context.add_message(
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": "Start by greeting the user and ask how you can help.",
                 }
             )

examples/foundational/23-bot-background-sound.py

@@ -128,7 +128,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         logger.info(f"Re-enabling background sound and starting bot...")
         await task.queue_frame(MixerEnableFrame(True))
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/24-user-mute-strategy.py

@@ -143,7 +143,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         # Kick off the conversation with a weather-related prompt
         context.add_message(
             {
-                "role": "system",
+                "role": "user",
                 "content": "Ask the user what city they'd like to know the weather for.",
             }
         )

examples/foundational/26b-gemini-live-function-calling.py

@@ -131,13 +131,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     # than as arguments to GeminiLiveLLMService, but note that doing so will
     # trigger a (fast) reconnection when the GeminiLiveLLMService first
     # receives the context (i.e. when we send the LLMRunFrame below).
-    context = LLMContext(
+    context = LLMContext()
-        [
-            # {"role": "system", "content": system_instruction},
-            {"role": "user", "content": "Say hello."},
-        ],
-        # tools,
-    )
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(
@@ -172,6 +166,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/28-user-assistant-turns.py

@@ -169,7 +169,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Start conversation - empty prompt to let LLM follow system instructions
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/29-turn-tracking-observer.py

@@ -191,7 +191,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/30-observer.py

@@ -158,7 +158,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/33-gemini-rag.py

@@ -197,11 +197,11 @@ Your response will be turned into speech so use only simple words and punctuatio
 """
 
     llm = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
         settings=GoogleLLMSettings(
             model=VOICE_MODEL,
             system_instruction=system_prompt,
         ),
-        api_key=os.getenv("GOOGLE_API_KEY"),
     )
     llm.register_function("query_knowledge_base", query_knowledge_base)
 
@@ -218,11 +218,7 @@ Your response will be turned into speech so use only simple words and punctuatio
     )
     tools = ToolsSchema(standard_tools=[query_function])
 
-    messages = [
+    context = LLMContext(tools=tools)
-        {"role": "user", "content": "Greet the user."},
-    ]
-
-    context = LLMContext(messages, tools)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
@@ -252,6 +248,7 @@ Your response will be turned into speech so use only simple words and punctuatio
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Start conversation - empty prompt to let LLM follow system instructions
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/38-smart-turn-fal.py

@@ -116,7 +116,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/38a-smart-turn-local-coreml.py

@@ -130,7 +130,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/38b-smart-turn-local.py

@@ -102,7 +102,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/39-mcp-stdio.py

@@ -142,7 +142,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             ),
         )
 
-        system = f"""
+        system_prompt = f"""
         You are a helpful LLM in a WebRTC call.
         Your goal is to demonstrate your capabilities in a succinct way.
         You have access to tools to search the Rijksmuseum collection.
@@ -158,7 +158,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         llm = AnthropicLLMService(
             api_key=os.getenv("ANTHROPIC_API_KEY"),
             settings=AnthropicLLMSettings(
-                system_instruction=system,
+                system_instruction=system_prompt,
             ),
         )
 

examples/foundational/39a-mcp-streamable-http.py

@@ -63,7 +63,19 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         ),
     )
 
-    llm = GoogleLLMService(api_key=os.getenv("GOOGLE_API_KEY"), model="gemini-2.0-flash")
+    system_prompt = f"""
+    You are a helpful LLM in a WebRTC call.
+    Your goal is to answer questions about the user's GitHub repositories and account.
+    You have access to a number of tools provided by Github. Use any and all tools to help users.
+    Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points.
+    Don't overexplain what you are doing.
+    Just respond with short sentences when you are carrying out tool calls.
+    """
+
+    llm = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
+        system_instruction=system_prompt,
+    )
 
     try:
         # Github MCP docs: https://github.com/github/github-mcp-server
@@ -87,18 +99,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         logger.error(f"error registering tools")
         logger.exception("error trace:")
 
-    system = f"""
+    context = LLMContext(tools=tools)
-    You are a helpful LLM in a WebRTC call.
-    Your goal is to answer questions about the user's GitHub repositories and account.
-    You have access to a number of tools provided by Github. Use any and all tools to help users.
-    Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points.
-    Don't overexplain what you are doing.
-    Just respond with short sentences when you are carrying out tool calls.
-    """
-
-    messages = [{"role": "system", "content": system}]
-
-    context = LLMContext(messages, tools)
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
         context,
         user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),

examples/foundational/39c-multiple-mcp.py

@@ -125,7 +125,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             ),
         )
 
-        system = f"""
+        system_prompt = f"""
         You are a helpful LLM in a WebRTC call.
         Your goal is to demonstrate your capabilities in a succinct way.
         You have access to tools to search the Rijksmuseum collection and the user's GitHub repositories and account.
@@ -142,7 +142,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         llm = AnthropicLLMService(
             api_key=os.getenv("ANTHROPIC_API_KEY"),
             settings=AnthropicLLMSettings(
-                system_instruction=system,
+                system_instruction=system_prompt,
             ),
         )
 

examples/foundational/40-aws-nova-sonic.py

@@ -146,15 +146,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     )
 
     # Set up context and context management.
-    context = LLMContext(
+    context = LLMContext(tools=tools)
-        messages=[
-            {
-                "role": "user",
-                "content": "Tell me a fun fact!",
-            },
-        ],
-        tools=tools,
-    )
     user_aggregator, assistant_aggregator = LLMContextAggregatorPair(context)
 
     # Build the pipeline
@@ -183,6 +175,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
         # HACK: if using the older Nova Sonic (pre-2) model, you need this special way of
         # triggering the first assistant response. Note that this trigger requires a special

examples/foundational/42-interruption-config.py

@@ -108,7 +108,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/43-heygen-transport.py

@@ -100,7 +100,7 @@ async def main():
             # Kick off the conversation.
             context.add_message(
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": "Start by saying 'Hello' and then a short greeting.",
                 }
             )

examples/foundational/43a-heygen-video-service.py

@@ -133,7 +133,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
             # Kick off the conversation.
             context.add_message(
                 {
-                    "role": "system",
+                    "role": "user",
                     "content": "Start by saying 'Hello' and then a short greeting.",
                 }
             )

examples/foundational/45-before-and-after-events.py

@@ -120,7 +120,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         # Custom frames are pushed in order so they can be used for synchronization purposes.
         await task.queue_frames([CustomBeforeProcessFrame(), LLMRunFrame(), CustomAfterPushFrame()])
 

examples/foundational/47-sentry-metrics.py

@@ -111,7 +111,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info(f"Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")

examples/foundational/48-service-switcher.py

@@ -116,15 +116,15 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
         services=[tts_cartesia, tts_deepgram], strategy_type=ServiceSwitcherStrategyManual
     )
 
-    system = "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way."
+    system_prompt = "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way."
 
     llm_openai = OpenAILLMService(
         api_key=os.getenv("OPENAI_API_KEY"),
-        settings=OpenAILLMSettings(system_instruction=system),
+        settings=OpenAILLMSettings(system_instruction=system_prompt),
     )
     llm_google = GoogleLLMService(
         api_key=os.getenv("GOOGLE_API_KEY"),
-        settings=GoogleLLMSettings(system_instruction=system),
+        settings=GoogleLLMSettings(system_instruction=system_prompt),
     )
     llm_switcher = LLMSwitcher(
         llms=[llm_openai, llm_google], strategy_type=ServiceSwitcherStrategyManual

examples/foundational/54-context-summarization-openai.py

@@ -175,7 +175,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     async def on_client_connected(transport, client):
         logger.info("Client connected")
         # Kick off the conversation.
-        context.add_message({"role": "system", "content": "Please introduce yourself to the user."})
+        context.add_message({"role": "user", "content": "Please introduce yourself to the user."})
         await task.queue_frames([LLMRunFrame()])
 
     @transport.event_handler("on_client_disconnected")