Vansh Kumar Singh Puts Google's Gemini AI in Your Pocket with This M5Stack Cardputer Gen AI Project
Linking to Gemini's API over the Cardputer's Wi-Fi connection, this dedicated pocket AI gadget is ready in an instant.
Self-described "tech enthusiast" Vansh Kumar Singh has turned an M5Stack Cardputer into a pocket-friendly artificial intelligence (AI) assistant linked to Google's Gemini multimodal model for on-the-go generative AI fun.
"[This is] an application that uses [M5Stack] Cardputer to interact with the Gemini API [Application Programming Interface] for generating responses based on user input," Singh explains of his work. "This project includes code for managing Wi-Fi connectivity, handling user input, making API requests, and displaying responses on the Cardputer's screen."
M5Stack launched the Cardputer in October last year as a pocket-sized portable development device featuring a cramped but usable physical keyboard and a full-color display, linked to an M5StampS3 microcontroller in a socket at the top-right. Gemini, meanwhile, is the new name for what was formerly known as Bard, a generative AI platform from Google designed to respond to natural language prompts with text and images.
Singh's project runs on the Cardputer and accepts these prompts through its built-in keyboard, using the integrated Wi-Fi radio in the Espressif ESP32-S3 chip at its heart to send the query off to Google's Gemini model via the official API. Once a response is received, it's processed for display on the Cardputer's compact 1.14" IPS display.
Singh has published the project on GitHub under the permissive MIT license, but notes that anyone interested in trying it it will need to create their own configuration file on the Cardputer's microSD card with their Wi-Fi access details and Gemini API key before it'll run.