Build a Voice-Controlled IoT Command Line with Telnyx

Voice AI demos often stop at conversation. A caller asks a question, an assistant replies, and nothing visible changes.

This example shows a more concrete workflow: call a phone number, say a facility command, and watch simulated IoT device state update in real time.

The app is a Python Flask service powered by Telnyx Voice AI and Call Control. It answers an inbound call, captures a spoken command, maps that command to a known device/action pair, updates in-memory device state, and speaks back a short confirmation.

What You Build

The demo behaves like a lightweight facility control line.

A caller can say:

unlock the loading dock gate
turn on the warehouse lights
check the freezer alarm
start the backup generator
turn off the irrigation pump

The app updates simulated devices like:

• loading_dock_gate
• warehouse_lights
• freezer_alarm
• backup_generator
• irrigation_pump

Each command is logged with the caller input, device, action, confidence, status, and response.

Why This Demo Works

The best voice ai demos make the output visible.

This example is easy to understand because the call changes state you can inspect through simple HTTP endpoints:

curl http://localhost:5000/devices
curl http://localhost:5000/commands

That makes the phone call feel connected to a real system instead of a standalone voice conversation.

Architecture

inbound phone call
  -> telnyx call control webhook
  -> flask app answers
  -> telnyx ai gather captures speech
  -> local parser maps device and action
  -> simulated iot state updates
  -> telnyx speaks confirmation
  -> command appears in audit log

The local API path can also use Telnyx AI Inference for command parsing, which is useful for showing how natural-language parsing can generalize beyond the deterministic demo phrases.

For the live phone demo, the app uses a fast local parser after AI gather captures the caller's speech. That keeps the call responsive and avoids an extra model round trip before speaking the result.

Run The Example

git clone https://github.com/team-telnyx/telnyx-code-examples.git
cd telnyx-code-examples/voice-activated-iot-command-python
cp .env.example .env
pip install -r requirements.txt
python app.py

Expose your local app:

ngrok http 5000

Set the Call Control webhook URL to:

https://<ngrok-domain>/webhooks/voice

Assign a Telnyx number to the Call Control application and call it.

Demo Script

Open the device state:

curl http://localhost:5000/devices

Then call the number and say:

unlock the loading dock gate

Refresh /devices and show that the gate is now unlocked.

Then say:

turn on the warehouse lights

Refresh /devices again and show the lights are on.

Finally, open:

curl http://localhost:5000/commands

That endpoint shows the command log, which is the beginning of a production audit trail.

Production Considerations

Before connecting this to real equipment, add:

• Caller authentication and allowlists
• Confirmation steps for risky actions
• Persistent command storage
• Real IoT device integrations
• Webhook signature verification
• Monitoring for failed call commands
• Rate limits and audit exports