How to Connect TFT LCD (ILI9341, ST7789) to an AI Agent: A Step-by-Step Guide with ASI Biont

Why Connect a TFT LCD Display to an AI Agent?

TFT LCD displays powered by controllers like ILI9341 and ST7789 are staples in IoT projects—from weather stations and smart home dashboards to industrial monitoring panels. Typically, you’d write custom firmware to update the screen with data from sensors or APIs. But what if you could offload all that logic to an AI agent that writes the code, manages the data, and even responds to your chat commands?

With ASI Biont, you can do exactly that. Instead of manually coding display updates, you describe what you want in plain English, and the AI agent writes the integration on the fly. This guide walks you through connecting an ESP32 with a TFT LCD (ILI9341 or ST7789) to ASI Biont via MQTT, so the AI can push real-time data—like sensor readings, alerts, or even simple UI elements—to your screen.

Which Connection Method and Why?

For TFT LCDs connected to ESP32 or Arduino, the most practical approach is MQTT (Message Queuing Telemetry Transport). Here’s why:

• Wireless freedom: ESP32 has built-in Wi-Fi, so no USB tethering needed.
• Two-way communication: The AI can both read sensor data and send display commands.
• Scalability: Multiple displays can subscribe to the same topics.

ASI Biont uses the paho-mqtt library inside its execute_python sandbox. The AI writes a Python script that connects to a broker (e.g., Mosquitto or HiveMQ Cloud), subscribes to a topic for incoming data, and publishes commands to a topic the ESP32 listens to.

For wired connections (e.g., Arduino Uno via USB), you’d use the Hardware Bridge method with bridge.py and COM port, but MQTT is generally cleaner for ESP32.

Step-by-Step: ESP32 + ILI9341 + ASI Biont via MQTT

What You Need

• ESP32 board (e.g., ESP32 DevKit V1)
• TFT LCD with ILI9341 driver (240×320 SPI) or ST7789 (240×240)
• DHT22 temperature/humidity sensor (optional, for sensor data)
• MQTT broker (e.g., HiveMQ Cloud free tier or local Mosquitto)
• ASI Biont account (free tier works)

Wiring (ILI9341 Example)

For ST7789, pins are similar; just adjust TFT_CS, TFT_DC, TFT_RST in code.

ESP32 Code (MicroPython)

Save the following as main.py on your ESP32 using Thonny or ampy:

import network
import time
from machine import Pin, SPI
import ili9341  # ili9341 library from https://github.com/jeffmer/micropython-ili9341

# Wi-Fi credentials
SSID = 'your_wifi'
PASSWORD = 'your_password'

# MQTT setup
import ubinascii
import umqtt.simple as mqtt

CLIENT_ID = ubinascii.hexlify(machine.unique_id())
BROKER = 'your_broker_address'
TOPIC_DISPLAY = 'home/display/command'
TOPIC_SENSOR = 'home/sensor/data'

# Display init
spi = SPI(2, baudrate=40000000, sck=Pin(18), mosi=Pin(23))
display = ili9341.ILI9341(spi, cs=Pin(5), dc=Pin(19), rst=Pin(18))
display.fill(ili9341.color565(0,0,0))
display.text('Connecting...', 10, 10, ili9341.color565(255,255,255))

# Wi-Fi connection
wlan = network.WLAN(network.STA_IF)
wlan.active(True)
wlan.connect(SSID, PASSWORD)
while not wlan.isconnected():
    time.sleep(1)
display.fill(ili9341.color565(0,0,0))
display.text('WiFi OK', 10, 10, ili9341.color565(0,255,0))

# MQTT callback
def sub_cb(topic, msg):
    msg_str = msg.decode()
    # Parse JSON command from AI
    import json
    try:
        cmd = json.loads(msg_str)
        if cmd['action'] == 'text':
            display.fill_rect(0, 30, 320, 30, ili9341.color565(0,0,0))
            display.text(cmd['data'], 10, 30, ili9341.color565(255,255,0))
        elif cmd['action'] == 'sensor':
            temp = cmd['temperature']
            hum = cmd['humidity']
            display.fill_rect(0, 70, 320, 40, ili9341.color565(0,0,0))
            display.text('Temp: {} C'.format(temp), 10, 70, ili9341.color565(0,255,0))
            display.text('Hum: {}%'.format(hum), 10, 90, ili9341.color565(0,255,0))
    except:
        display.fill_rect(0, 30, 320, 30, ili9341.color565(0,0,0))
        display.text(msg_str, 10, 30, ili9341.color565(255,0,0))

client = mqtt.MQTTClient(CLIENT_ID, BROKER)
client.set_callback(sub_cb)
client.connect()
client.subscribe(TOPIC_DISPLAY)

# Main loop
while True:
    client.check_msg()
    time.sleep(0.1)

How ASI Biont Connects

You tell ASI Biont in chat: “Connect to my MQTT broker at broker.hivemq.cloud:1883, subscribe to home/sensor/data, and publish display commands to home/display/command. The ESP32 has a DHT22 sensor and an ILI9341 screen. Every 30 minutes, read temperature and humidity, show them on the display, and if temperature exceeds 30°C, publish an alert to the display topic.”

ASI Biont then writes and runs this Python script in its sandbox:

import paho.mqtt.client as mqtt
import json
import time
import random  # simulate sensor; in real use, read via MQTT from ESP32

BROKER = 'broker.hivemq.cloud'
PORT = 1883
TOPIC_SENSOR = 'home/sensor/data'
TOPIC_DISPLAY = 'home/display/command'

client = mqtt.Client()
client.connect(BROKER, PORT, 60)

# Simulate sensor data (replace with actual subscription)
for _ in range(3):  # limited to 3 cycles due to sandbox timeout
    temp = round(20 + random.uniform(-5, 15), 1)
    hum = round(40 + random.uniform(-10, 20), 1)

    # Publish sensor data
    client.publish(TOPIC_SENSOR, json.dumps({'temperature': temp, 'humidity': hum}))

    # Check threshold
    if temp > 30:
        alert = {'action': 'text', 'data': 'ALERT: High temperature!'}
        client.publish(TOPIC_DISPLAY, json.dumps(alert))
    else:
        display_cmd = {'action': 'sensor', 'temperature': temp, 'humidity': hum}
        client.publish(TOPIC_DISPLAY, json.dumps(display_cmd))

    time.sleep(10)

client.disconnect()

The AI can also be asked to generate a live dashboard: “Show a bar chart of temperature over the last hour on the display.” It would then publish pixel data or simple shape commands.

Real-World Scenarios

1. Telegram + Display: Ask ASI Biont: “When I send ‘show temp’ via Telegram, publish the current temperature to the display.” The AI listens to Telegram webhooks and forwards data.
2. Predictive Alerts: “If temperature rises faster than 2°C per minute, show a red warning on the screen.” The AI computes rate of change from MQTT stream.
3. Remote Control: “Let me turn on an LED connected to ESP32 by typing ‘led on’ in chat.” The AI publishes a command to a topic the ESP32 subscribes to.

Why This Matters

You don’t need to write complex firmware or learn MQTT deeply. ASI Biont acts as your personal IoT developer—it writes the Python code, handles the MQTT logic, and updates your display based on natural language commands. No dashboards to configure, no “add device” buttons. Just describe what you want, and the AI does the rest.

For wired devices (e.g., Arduino Uno with ST7789), the same principle applies via Hardware Bridge—the AI uses industrial_command with serial:// protocol to send data over USB. The chat handles everything.

Pitfalls to Avoid

• Sandbox timeout: ASI Biont’s execute_python has a 30-second limit. For long-running tasks, use MQTT subscriptions that trigger on events, not infinite loops.
• MQTT QoS: Use QoS 0 for display commands (fire-and-forget), but QoS 1 for critical sensor alerts.
• Display flicker: Clear only changed areas (fill_rect) instead of full screen to avoid flicker.
• Power: ESP32 with TFT can draw ~80mA; use deep sleep between updates to save battery.

Try It Yourself

1. Sign up at asibiont.com (free tier).
2. Wire your ESP32 + TFT LCD as described.
3. In the chat, type: “Connect to my MQTT broker at [your broker] and run a script that every 5 minutes reads a simulated sensor and shows it on my ILI9341 display.”
4. Watch the AI write and execute the code in seconds.

From simple text notifications to dynamic dashboards, ASI Biont turns your TFT display into a smart, AI-controlled interface. Start integrating today.