diff --git a/.github/workflows/poetry-test.yml b/.github/workflows/poetry-test.yml
index 96dfc8e45..1f01701a9 100644
--- a/.github/workflows/poetry-test.yml
+++ b/.github/workflows/poetry-test.yml
@@ -38,6 +38,12 @@ jobs:
       - name: Install Poetry
         uses: snok/install-poetry@v1
 
+      - name: Create virtual audio device
+        run: |
+          apt-get update
+          DEBIAN_FRONTEND=noninteractive apt-get --yes install jackd
+          jackd -d dummy -r 44100 &
+
       - name: Install python dependencies
         run: poetry install --with openset,nomad
 
@@ -63,4 +69,4 @@ jobs:
         run: |
           source /opt/ros/${{ matrix.ros_distro }}/setup.bash
           source install/setup.bash
-          poetry run pytest
+          poetry run pytest -m "not billable"
diff --git a/pyproject.toml b/pyproject.toml
index 20e7eb32f..4457ea4cb 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -85,7 +85,8 @@ profile = "black"
 [tool.pytest.ini_options]
 markers = [
     "billable: marks test as billable  (deselect with '-m \"not billable\"')",
+    "ci_only: marks test as cli only (deselect with '-m \"not ci_only\"')",
 ]
-addopts = "-m 'not billable' --ignore=src"
+addopts = "-m 'not billable and not ci_only' --ignore=src"
 log_cli = true
 log_cli_level = "DEBUG"
diff --git a/src/rai/rai/agents/__init__.py b/src/rai/rai/agents/__init__.py
index ef74fc891..dc101282b 100644
--- a/src/rai/rai/agents/__init__.py
+++ b/src/rai/rai/agents/__init__.py
@@ -11,3 +11,13 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
+from rai.agents.conversational_agent import create_conversational_agent
+from rai.agents.state_based import create_state_based_agent
+from rai.agents.tool_runner import ToolRunner
+
+__all__ = [
+    "ToolRunner",
+    "create_conversational_agent",
+    "create_state_based_agent",
+]
diff --git a/src/rai/rai/communication/__init__.py b/src/rai/rai/communication/__init__.py
new file mode 100644
index 000000000..f22b87447
--- /dev/null
+++ b/src/rai/rai/communication/__init__.py
@@ -0,0 +1,23 @@
+# Copyright (C) 2024 Robotec.AI
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from .base_connector import BaseConnector, BaseMessage
+from .sound_device_connector import SoundDeviceError, StreamingAudioInputDevice
+
+__all__ = [
+    "BaseMessage",
+    "BaseConnector",
+    "StreamingAudioInputDevice",
+    "SoundDeviceError",
+]
diff --git a/src/rai/rai/communication/base_connector.py b/src/rai/rai/communication/base_connector.py
new file mode 100644
index 000000000..fe01097fc
--- /dev/null
+++ b/src/rai/rai/communication/base_connector.py
@@ -0,0 +1,49 @@
+# Copyright (C) 2024 Robotec.AI
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from abc import ABC, abstractmethod
+from typing import Callable
+from uuid import uuid4
+
+
+class BaseMessage(ABC):
+    pass
+
+
+class BaseConnector(ABC):
+
+    def _generate_handle(self) -> str:
+        return str(uuid4())
+
+    @abstractmethod
+    def send_message(self, msg: BaseMessage, target: str) -> None:
+        pass
+
+    @abstractmethod
+    def receive_message(self, source: str) -> BaseMessage:
+        pass
+
+    @abstractmethod
+    def send_and_wait(self, target: str) -> BaseMessage:
+        pass
+
+    @abstractmethod
+    def start_action(
+        self, target: str, on_feedback: Callable, on_finish: Callable = lambda _: None
+    ) -> str:
+        pass
+
+    @abstractmethod
+    def terminate_action(self, action_handle: str):
+        pass
diff --git a/src/rai/rai/communication/sound_device_connector.py b/src/rai/rai/communication/sound_device_connector.py
new file mode 100644
index 000000000..f88619c2b
--- /dev/null
+++ b/src/rai/rai/communication/sound_device_connector.py
@@ -0,0 +1,141 @@
+# Copyright (C) 2024 Robotec.AI
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from typing import Any, Callable, Optional, TypedDict
+
+import numpy as np
+import sounddevice as sd
+from scipy.signal import resample
+from sounddevice import CallbackFlags
+
+from rai.communication.base_connector import BaseConnector, BaseMessage
+
+
+class SoundDeviceError(Exception):
+    def __init__(self, msg: str):
+        super().__init__(msg)
+
+
+class AudioInputDeviceConfig(TypedDict):
+    block_size: int
+    consumer_sampling_rate: int
+    target_sampling_rate: int
+    dtype: str
+    device_number: Optional[int]
+
+
+class ConfiguredAudioInputDevice:
+    """
+    A class to store the configuration of an audio device
+
+    Attributes
+    ----------
+    sample_rate (int): Device sample rate
+    consumer_sampling_rate (int): The sampling rate of the consumer
+    window_size_samples (int): The size of the window in samples
+    target_sampling_rate (int): The target sampling rate
+    dtype (str): The data type of the audio samples
+    """
+
+    def __init__(self, config: AudioInputDeviceConfig):
+        self.sample_rate = sd.query_devices(
+            device=config["device_number"], kind="input"
+        )[
+            "default_samplerate"
+        ]  # type: ignore
+        self.consumer_sampling_rate = config["consumer_sampling_rate"]
+        self.window_size_samples = int(
+            config["block_size"] * self.sample_rate / config["consumer_sampling_rate"]
+        )
+        self.target_sampling_rate = int(config["target_sampling_rate"])
+        self.dtype = config["dtype"]
+
+
+class StreamingAudioInputDevice(BaseConnector):
+    def __init__(self):
+        self.streams = {}
+        sd.default.latency = ("low", "low")
+        self.configred_devices: dict[str, ConfiguredAudioInputDevice] = {}
+
+    def configure_device(self, target: str, config: AudioInputDeviceConfig):
+        if target.isdigit():
+            if config.get("device_number") is None:
+                config["device_number"] = int(target)
+            elif config["device_number"] != int(target):
+                raise SoundDeviceError(
+                    "device_number in config must be the same as target"
+                )
+            self.configred_devices[target] = ConfiguredAudioInputDevice(config)
+        else:
+            raise SoundDeviceError("target must be a device number!")
+
+    def send_message(self, msg: BaseMessage, target: str) -> None:
+        raise SoundDeviceError(
+            "StreamingAudioInputDevice does not suport sending messages"
+        )
+
+    def receive_message(self, source: str) -> BaseMessage:
+        raise SoundDeviceError(
+            "StreamingAudioInputDevice does not suport receiving messages messages"
+        )
+
+    def send_and_wait(self, target: str) -> BaseMessage:
+        raise SoundDeviceError(
+            "StreamingAudioInputDevice does not suport sending messages"
+        )
+
+    def start_action(
+        self,
+        target: str,
+        on_feedback: Callable[[np.ndarray, dict[str, Any]], None],
+        on_finish: Callable = lambda _: None,
+    ) -> str:
+
+        target_device = self.configred_devices.get(target)
+        if target_device is None:
+            raise SoundDeviceError(f"Device {target} has not been configured")
+
+        def callback(indata: np.ndarray, frames: int, _, status: CallbackFlags):
+            indata = indata.flatten()
+            sample_time_length = len(indata) / target_device.target_sampling_rate
+            if target_device.sample_rate != target_device.target_sampling_rate:
+                indata = resample(indata, int(sample_time_length * target_device.target_sampling_rate))  # type: ignore
+            flag_dict = {
+                "input_overflow": status.input_overflow,
+                "input_underflow": status.input_underflow,
+                "output_overflow": status.output_overflow,
+                "output_underflow": status.output_underflow,
+                "priming_output": status.priming_output,
+            }
+            on_feedback(indata, flag_dict)
+
+        handle = self._generate_handle()
+        try:
+            stream = sd.InputStream(
+                samplerate=target_device.sample_rate,
+                channels=1,
+                device=int(target),
+                dtype=target_device.dtype,
+                blocksize=target_device.window_size_samples,
+                callback=callback,
+                finished_callback=on_finish,
+            )
+        except AttributeError:
+            raise SoundDeviceError(f"Device {target} has not been correctly configured")
+        stream.start()
+        self.streams[handle] = stream
+        return handle
+
+    def terminate_action(self, action_handle: str):
+        self.streams[action_handle].stop()
diff --git a/tests/communication/test_sound_device_connector.py b/tests/communication/test_sound_device_connector.py
new file mode 100644
index 000000000..cb6d72101
--- /dev/null
+++ b/tests/communication/test_sound_device_connector.py
@@ -0,0 +1,114 @@
+# Copyright (C) 2024 Robotec.AI
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#         http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from unittest import mock
+
+import pytest
+import sounddevice as sd
+
+from rai.communication import SoundDeviceError, StreamingAudioInputDevice
+
+
+@pytest.fixture
+def setup_mock_input_stream():
+    with mock.patch("sounddevice.InputStream") as mock_input_stream:
+        yield mock_input_stream
+
+
+@pytest.fixture
+def device_config():
+    return {
+        "block_size": 1024,
+        "consumer_sampling_rate": 44100,
+        "target_sampling_rate": 16000,
+        "dtype": "float32",
+    }
+
+
+@pytest.mark.ci_only
+def test_configure(
+    setup_mock_input_stream,
+    device_config,
+):
+    mock_input_stream = setup_mock_input_stream
+    mock_instance = mock.MagicMock()
+    mock_input_stream.return_value = mock_instance
+    audio_input_device = StreamingAudioInputDevice()
+    device = sd.query_devices(kind="input")
+    if type(device) is dict:
+        device_id = str(device["index"])
+    elif isinstance(device, list):
+        device_id = str(device[0]["index"])  # type: ignore
+    else:
+        raise AssertionError("No input device found")
+    audio_input_device.configure_device(device_id, device_config)
+    assert (
+        audio_input_device.configred_devices[device_id].consumer_sampling_rate == 44100
+    )
+    assert audio_input_device.configred_devices[device_id].window_size_samples == 1024
+    assert audio_input_device.configred_devices[device_id].target_sampling_rate == 16000
+    assert audio_input_device.configred_devices[device_id].dtype == "float32"
+
+
+@pytest.mark.ci_only
+def test_start_action_failed_init(
+    setup_mock_input_stream,
+):
+    mock_input_stream = setup_mock_input_stream
+    mock_instance = mock.MagicMock()
+    mock_input_stream.return_value = mock_instance
+    audio_input_device = StreamingAudioInputDevice()
+
+    feedback_callback = mock.MagicMock()
+    finish_callback = mock.MagicMock()
+
+    recording_device = 0
+    with pytest.raises(SoundDeviceError, match="Device 0 has not been configured"):
+        _ = audio_input_device.start_action(
+            str(recording_device), feedback_callback, finish_callback
+        )
+
+
+@pytest.mark.ci_only
+def test_start_action(
+    setup_mock_input_stream,
+    device_config,
+):
+    mock_input_stream = setup_mock_input_stream
+    mock_instance = mock.MagicMock()
+    mock_input_stream.return_value = mock_instance
+    audio_input_device = StreamingAudioInputDevice()
+
+    feedback_callback = mock.MagicMock()
+    finish_callback = mock.MagicMock()
+
+    device = sd.query_devices(kind="input")
+    if type(device) is dict:
+        device_id = str(device["index"])
+    elif isinstance(device, list):
+        device_id = str(device[0]["index"])  # type: ignore
+    else:
+        raise AssertionError("No input device found")
+    audio_input_device.configure_device(device_id, device_config)
+
+    stream_handle = audio_input_device.start_action(
+        device_id, feedback_callback, finish_callback
+    )
+
+    assert mock_input_stream.call_count == 1
+    init_args = mock_input_stream.call_args.kwargs
+    assert init_args["device"] == int(device_id)
+    assert init_args["finished_callback"] == finish_callback
+
+    assert audio_input_device.streams.get(stream_handle) is not None