Build a Pocket AI Chatbot on Raspberry Pi Zero 2 W

Flash Raspberry Pi OS and enable remote access

Use Raspberry Pi Imager to write current 64-bit Raspberry Pi OS. Set the hostname, username, password, Wi-Fi and SSH before the first boot.

• Use a 16 GB or larger microSD card.
• Record the hostname and login.
• Keep Wi-Fi available because this version requires cloud access during use.

Raspberry Pi Imager →Optional PiSugar prebuilt image →

Assemble the battery, Pi Zero and Whisplay HAT

With all power disconnected, mount the PiSugar 3 beneath the Raspberry Pi Zero 2 W and align the Whisplay HAT over the complete 40-pin header.

• Never attach or move a HAT while powered.
• Check the first and last GPIO pins before pressing the boards together.
• Use the supplied spacers so no solder joint or metal fastener touches the battery.

Boot, connect over SSH and update the Pi

Start on stable external power, connect to the configured hostname and update Raspberry Pi OS before adding the hardware driver.

ssh your-user@your-hostname.local
sudo apt update
sudo apt full-upgrade -y

Install and test the reviewed Whisplay driver

Clone the official driver, pin the revision reviewed by LocalClaw, install it and reboot. Run the supplied hardware example before installing the chatbot.

cd ~
git clone https://github.com/PiSugar/Whisplay.git --depth 1
cd Whisplay
git checkout a695240f866257033414697d47c3aa4c5f9b749c
sudo bash install_driver.sh
sudo reboot

Note: After reboot, use the official example test to confirm the display, button, LEDs, microphone and speaker. Do not debug the cloud API until the hardware passes.

Install and build the reviewed chatbot

Clone the PiSugar application, pin the reviewed revision, install its dependencies and build the interface.

cd ~
git clone https://github.com/PiSugar/whisplay-ai-chatbot.git
cd whisplay-ai-chatbot
git checkout ee7301b2f7693111ee016f81c48a75d3387a7326
bash install_dependencies.sh
source ~/.bashrc
bash build.sh

Configure cloud speech, chat and voice safely

Run the interactive wizard and select OpenAI for ASR, LLM and TTS. Enter the API key only on the Raspberry Pi, restrict the .env file and verify provider names without printing the secret.

• ASR_SERVER=openai
• LLM_SERVER=openai
• TTS_SERVER=openai
• Never paste OPENAI_API_KEY into screenshots, shell history, Git or support messages.
• Review current API pricing and set account limits before repeated tests.

cd ~/whisplay-ai-chatbot
whisplay configure
chmod 600 .env
grep -E '^(ASR_SERVER|LLM_SERVER|TTS_SERVER)=' .env

Note: The reviewed source currently defaults to hosted models when explicit overrides are absent. Provider availability and pricing can change, so confirm the current OpenAI documentation before choosing overrides.

Run one complete push-to-talk test

Start the chatbot in the foreground, press the Whisplay button, speak a short request and wait for the displayed and spoken answer.

cd ~/whisplay-ai-chatbot
bash run_chatbot.sh

Note: A successful answer proves hardware, network and API access together. If volume is low, open alsamixer and adjust the Whisplay output before changing AI settings.

Install the PiSugar manager and check the battery

Install the official PiSugar power manager, select the PiSugar 3 model when prompted and verify that its service can read the battery.

cd ~
wget -O pisugar-power-manager.sh https://cdn.pisugar.com/release/pisugar-power-manager.sh
bash pisugar-power-manager.sh -c release
systemctl status pisugar-server --no-pager

Note: The manager exposes a local web interface on port 8421. Change default credentials and do not expose it directly to the public internet.

Enable startup only after the foreground test passes

Stop the foreground process, run the official startup helper and inspect the service and log after reboot.

• Run startup.sh as the normal user, not with sudo.
• The helper may switch the Pi to headless multi-user mode.
• If whisplay-daemon is installed, follow the current repository warning instead of creating a competing service.

cd ~/whisplay-ai-chatbot
bash startup.sh
systemctl status chatbot --no-pager
tail -n 100 chatbot.log