Most smart displays want your attention. They light up, alert you, recommend things, and keep a steady stream of information moving across the room. Google’s newly open-sourced Glanceboard takes a noticeably different route: it turns a family schedule into a single, AI-generated image on an e-ink display, designed to be checked briefly rather than constantly consumed.

The project was built by Google creative technologist Raph Dixon and is now available as open-source code on GitHub. At its core, Glanceboard combines a calendar feed, local weather, and customizable family characters into an illustrated daily planner. The result is less like a dashboard full of widgets and more like a personalized picture frame that happens to tell you what the day looks like.

Each morning, the system reads calendar events through an iCal feed, retrieves local weather conditions, and uses Gemini to turn that information into a more natural, human-friendly plan. Google’s Nano Banana image-generation model then produces an illustration: family members might be shown in clothes suited to the forecast, while the image includes reminders about school, appointments, bags to pack, or other events.

That idea sounds whimsical, but it addresses a very ordinary household problem. Family schedules tend to live inside phones, apps, inboxes, and calendars that require someone to unlock a device and parse a wall of information. Glanceboard puts the day’s essentials in a shared physical space, without treating the home like another notification center.

A different kind of AI app

Glanceboard is also an interesting example of where the current “vibe coding” conversation is heading. The term broadly describes building software by directing an AI coding assistant in natural language rather than writing every component manually. According to the project’s repository, the demo itself was “entirely vibe coded” using Google Antigravity, with off-the-shelf hardware rather than a bespoke prototype.

That matters because Glanceboard is not being presented as a polished, mass-market Google hardware product. The repository explicitly says it is not an official Google product. Instead, it is a public blueprint: a working project that developers, hobbyists, and technically inclined households can inspect, modify, and build on.

In practice, that means the code is as much the announcement as the device. Google is showing what becomes possible when generative AI is paired with simple home hardware, familiar personal data such as a calendar, and a narrowly defined daily use case. It is an AI application with a clear job: make the morning easier to understand at a glance.

How Glanceboard works

The technical flow is relatively straightforward, even if the finished experience feels more magical than its components suggest. A lightweight local server runs on a computer or Raspberry Pi, periodically checking the user’s calendar and weather data. It creates a prompt, generates a new planner image, converts that image for a six-color e-ink panel, and serves it to the display. The display polls the local server for updates rather than relying on a permanent cloud account.

Glanceboard uses Gemini Flash 3.6 for the language and planning layer, including rewriting sparse calendar labels into friendlier reminders, generating weather-aware descriptions, and identifying countdown-worthy events such as trips, birthdays, or holidays. For the visual side, the project recommends Gemini’s Nano Banana Pro for image quality, text rendering, and keeping recurring family characters visually consistent.

The distinction is important. The app is not merely pasting calendar text onto an e-ink screen. It is interpreting the context around that text and presenting it in a deliberately softer format. A terse event such as “9 am Tavi Library Bag,” for example, can be converted into a more useful reminder, while the generated scene can reflect the day’s actual conditions.

The open-source version includes several optional extras: weather, stock tickers, sports scores, daily quotes, news, smart countdowns, and an unread-email digest. It also supports six art styles, including watercolor, comic-book, pixel-art, pen-and-ink, fashion sketch, and Japanese sumi-e. Users can add reference photos for people and pets, giving the image generator a better chance of producing recurring characters that feel connected to the household.

The hardware is intentionally ordinary

Rather than requiring a custom-built gadget, Glanceboard’s primary setup uses a 7.3-inch Waveshare ESP32-S3 PhotoPainter color e-ink display in a wooden frame, a USB-C power source, and a computer or Raspberry Pi to host the local server. The project says no soldering or ribbon cables are required for the standard setup.

That lowers the barrier considerably, although it is still a project for people comfortable setting up software, configuring Wi-Fi, and managing API keys. The software stack requires Python 3.9 or later for the server and Node.js 18 or later to build the web dashboard. Users also need a Gemini API key, an iCal URL for their calendar, and a location setting for weather information.

There is a small but meaningful irony here: the device aims to reduce screen dependence, yet its setup belongs firmly to the modern maker ecosystem of repositories, API credentials, local servers, and browser-based flashing tools. For many users, that will be part of the appeal. For others, it will underline why Glanceboard remains a concept-driven open-source project rather than an appliance ready for a retail shelf.

Privacy is a selling point – with caveats

Google is framing the local setup as a key part of Glanceboard’s appeal. The display itself fetches generated images from a local server, while configuration data and generated images are kept on the user’s own machine in the self-hosted setup.

Still, “local” does not mean the AI workflow is completely offline. Glanceboard requires a Gemini API key for image generation, and its AI features call Gemini models. Users should therefore understand the split: the server and display can run locally, but content sent to the chosen AI provider is subject to that provider’s terms and data handling practices.

The optional Gmail feature is a good example of the project’s boundaries. It requires users to create their own Google Cloud OAuth credentials, reads sender names and email subject lines rather than message bodies, and uses Gemini to create a short digest. Google describes it as optional, so the core calendar-and-weather experience does not depend on connecting Gmail.

For a household display that could surface children’s schedules, travel plans, and personal routines, those details are not minor footnotes. The more useful Glanceboard becomes, the more sensitive the information it can summarize. Open-sourcing the project gives users more visibility and control than a sealed smart-display product would, but it also places responsibility on them to configure it thoughtfully.

Why the e-ink format fits

The most compelling decision in Glanceboard may be the least flashy one: e-ink. Unlike a tablet, the display is not built for feeds, videos, or interaction. It is there to communicate a stable, low-distraction snapshot of the day.

That changes the emotional character of the product. A phone asks to be held. A smart display often asks to be talked to. An e-ink board can sit quietly in a kitchen or hallway, offering information without becoming another source of ambient digital noise.

Google’s own description leans into this point, calling Glanceboard a way to avoid bright screens and stressful notifications. The result is meant to help families understand schedules, clothing choices, and items to pack without turning every morning into a scramble across multiple devices.

There is also a practical advantage: e-ink does not need fast refresh rates for a once-or-twice-daily planner. Its limitations – slower updates and muted color – become part of the product’s design logic. The display is not trying to compete with a tablet; it is trying to replace the paper note on the refrigerator, only with a little more awareness of the day ahead.

More blueprint than product

Glanceboard arrives at a moment when AI companies are eager to show that generative models can do more than answer questions in chat windows. The project offers a more tangible demonstration: AI can act as a translation layer between messy digital information and the physical environments where people actually live.

Its value is not that it produces a pretty image of someone wearing a raincoat. Its value is that it reframes a shared calendar as something more approachable, especially for children or busy families who may not naturally check a scheduling app. A visual prompt to bring a library bag can land differently from another line of text in a phone notification.

The bigger question is whether projects like Glanceboard stay in the maker community or point toward a new category of consumer hardware. Google has not announced a standalone Glanceboard product, and the repository makes clear that the current version is a demo rather than an official Google offering.

But the concept is easy to imagine moving beyond its current form. A less technical version could eventually become a family display, a workplace status board, a care tool for older adults, or a school-day planner that makes routines easier to follow. The open-source release lets that experimentation begin now, in public, rather than waiting for a polished commercial product.

For the moment, Glanceboard is a clever, unusually human example of AI-assisted home computing: not another app demanding attention, but a quiet object that turns the day into something you can simply glance at.