diff --git a/examples/foundational/19a-tools-anthropic.py b/examples/foundational/19a-tools-anthropic.py
index 8e0fc6bbc..e238de63c 100644
--- a/examples/foundational/19a-tools-anthropic.py
+++ b/examples/foundational/19a-tools-anthropic.py
@@ -62,7 +62,7 @@ async def main():
         )
 
         llm = AnthropicLLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
+            api_key=os.getenv("ANTHROPIC_API_KEY"),
             model="claude-3-5-sonnet-20240620"
         )
         llm.register_function("get_weather", get_weather)
@@ -86,10 +86,12 @@ async def main():
 
         # todo: test with very short initial user message
 
-        messages = [{"role": "system",
-                     "content": "You are a helpful assistant who can report the weather in any location in the universe. Respond concisely. Your response will be turned into speech so use only simple words and punctuation."},
-                    {"role": "user",
-                     "content": " Start the conversation by introducing yourself."}]
+        # messages = [{"role": "system",
+        #              "content": "You are a helpful assistant who can report the weather in any location in the universe. Respond concisely. Your response will be turned into speech so use only simple words and punctuation."},
+        #             {"role": "user",
+        #              "content": " Start the conversation by introducing yourself."}]
+
+        messages = [{"role": "user", "content": "Say 'hello' to start the conversation."}]
 
         context = OpenAILLMContext(messages, tools)
         context_aggregator = llm.create_context_aggregator(context)
@@ -109,7 +111,7 @@ async def main():
         async def on_first_participant_joined(transport, participant):
             transport.capture_participant_transcription(participant["id"])
             # Kick off the conversation.
-            await task.queue_frames([LLMMessagesFrame(messages)])
+            await task.queue_frames([context_aggregator.user().get_context_frame()])
 
         runner = PipelineRunner()
 
diff --git a/examples/foundational/19b-tools-video-anthropic.py b/examples/foundational/19b-tools-video-anthropic.py
index 51a321bf9..4ba29ab37 100644
--- a/examples/foundational/19b-tools-video-anthropic.py
+++ b/examples/foundational/19b-tools-video-anthropic.py
@@ -137,7 +137,8 @@ If you need to use a tool, simply use the tool. Do not tell the user the tool yo
         """
 
         messages = [{"role": "system",
-                     "content": system_prompt,
+                     "content": system_prompt},
+                    {"role": "user",
                      "content": "Start the conversation by introducing yourself."}]
 
         context = OpenAILLMContext(messages, tools)
@@ -161,7 +162,7 @@ If you need to use a tool, simply use the tool. Do not tell the user the tool yo
             transport.capture_participant_transcription(video_participant_id)
             transport.capture_participant_video(video_participant_id, framerate=0)
             # Kick off the conversation.
-            await task.queue_frames([LLMMessagesFrame(messages)])
+            await task.queue_frames([context_aggregator.user().get_context_frame()])
 
         runner = PipelineRunner()
         await runner.run(task)
diff --git a/src/pipecat/services/anthropic.py b/src/pipecat/services/anthropic.py
index ec19164c1..506d71243 100644
--- a/src/pipecat/services/anthropic.py
+++ b/src/pipecat/services/anthropic.py
@@ -110,7 +110,7 @@ class AnthropicLLMService(LLMService):
             await self.start_ttfb_metrics()
 
             response = await self._client.messages.create(
-                system=context.system,
+                system=context.system or [],
                 messages=messages,
                 tools=context.tools or [],
                 model=self._model,
@@ -255,7 +255,9 @@ class AnthropicLLMContext(OpenAILLMContext):
 
     @classmethod
     def from_messages(cls, messages: List[dict]) -> "AnthropicLLMContext":
-        return cls(messages=messages)
+        self = cls(messages=messages)
+        self._restructure_from_openai_messages()
+        return self
 
     @classmethod
     def from_image_frame(cls, frame: VisionImageRawFrame) -> "AnthropicLLMContext":