Gemini Genie brings the power of Google's Gemini AI directly into your HomeGenie smart home. Directly from the AI chat, you can talk with your home naturally: ask how things are running, control your devices, tweak configurations, write new automation programs, and design dashboard widgets.
Because the AI runs on Google's cloud servers, your HomeGenie server stays light and responsive without needing a dedicated graphics card or local language models. All you need is an internet connection and a Gemini API key. HomeGenie manages the connection, passes only the context you choose, and executes your automations safely on your local network.
One assistant for your entire smart home
Think of an AI Genie as a capable co-pilot for your HomeGenie system. While Lailama lets you run models locally on your own computer, Gemini Genie connects to Google's cloud models for high-speed answers, advanced coding, and deep reasoning.
You never have to switch between different apps or menus. Within the same chat, you can move effortlessly from simple everyday checks to designing complete solutions. For example:
- “Are any exterior doors open, and which lights are currently turned on?”
- “Turn off all downstairs lights and set the porch light to 40%.”
- “Which program controls the motion sensor for the porch light, and what is its current shut-off delay?”
- “Change that shut-off delay to two minutes.”
- “Create an automation that notifies me if the garage door stays open for more than ten minutes, and build a widget to display its status.”
You always decide what the assistant can see and what it is allowed to do through your context and capability settings.
Setting up Gemini Genie
Getting started takes only a moment. Once the Gemini Genie program is installed and enabled through System settings, you can configure it in a few simple steps:
- Open the AI chat and select Gemini Genie.
- Click the gear icon to open the assistant settings (or open them directly from System settings).
- Enter your Google Gemini API key. If you need a key, you can create one in Google AI Studio.
- Select your preferred Model ID. The default
gemini-flash-latestmodel delivers an ideal mix of speed and intelligence, but you can choose any Gemini model supported by your account. - Save your settings, and you are ready to start.
Want to customize how your assistant communicates? Use the System instructions field to guide its tone and behavior. HomeGenie automatically provides the necessary instructions for any capabilities you enable, and you can add #use-skill or #use-kb directives to give the assistant specialized knowledge.
You control privacy and permissions
The toolbar at the top of the chat gives you complete control through two clear menus: Context (what the assistant knows) and Capabilities (what the assistant can do). HomeGenie remembers your preferences for each Genie.
Context: what information to share
| Category | Information available to the assistant |
|---|---|
| Devices | Device names, addresses, types, and real-time status, with feature details available on demand. |
| Automation Programs | Descriptions of your programs and their source code when relevant to the task. |
| Widgets | Available dashboard widgets along with their HTML, CSS, and JavaScript source code. |
| Dashboards | Dashboard layouts and associations between modules and widgets. |
| Deep Knowledge | HomeGenie technical reference materials and development guides. |
HomeGenie is designed to protect your privacy and bandwidth: it only retrieves source files and specific parameter values when your request actually requires them. Selecting a category does not send your entire system configuration with every message.
Keep in mind that your prompt and selected context are processed on Google's servers. If you adjust your context settings, the change applies to future messages without altering data already sent.
Capabilities: what the assistant may do
| Capability | Actions it enables |
|---|---|
| Device control | Controlling devices and applying configuration changes. |
| Automation | Creating, updating, and running automation programs. |
| Widgets | Generating and editing custom dashboard widgets. |
Information and actions are intentionally separated. For example, if you enable the Devices context but keep Device control turned off, your Genie can inspect a light and read its brightness, but cannot turn it on or off.
HomeGenie checks these permissions before running any action. If a request needs an option you have not enabled, the Genie can explain which option to enable in your settings. It cannot enable capabilities on its own.
Control devices with natural language
You do not need to memorize exact system IDs or technical addresses. Gemini recognizes everyday names like “porch light” or “hallway sensor”.
When you ask about device behaviors, the Genie looks behind the scenes: it discovers the features on that module, finds the program that controls them, and checks the exact values in question. When you ask to change a setting—such as an auto-off timer or a light level—HomeGenie validates supported configuration changes against the device feature definition and reports the execution result.
You can inspect device commands in the chat through the API command blocks and their execution results. You can expand Show code at any time to see the underlying command and copy it into your own scripts. These blocks serve as a clear history of what happened; reopening a past chat will never trigger those commands again.
Create programs with an AI coding partner
Turn your automation ideas into working code simply by describing what you want. Gemini Genie can generate complete HomeGenie programs in C#, JavaScript, or Python.
When Automation is enabled, a generation request can install and run the program right away. For the best results, tell the assistant how the program should operate: should it run once, run continuously in the background, or wait for an event trigger?
You can refine programs collaboratively over time. The Genie inspects the current code before making updates, keeping the program ID intact. If a new program encounters a compilation issue, HomeGenie automatically reports the error back to the assistant so it can attempt a correction. Some issues may still require your review or further troubleshooting.
A Completed status signals that a task has finished; it does not automatically stop or disable the program. A Quit status stops and disables the program while preserving its source code, so you can enable it again later.
Design custom widgets and full solutions
Need a specialized interface for your wall tablet or phone? Ask for a custom widget, and Gemini Genie will generate the HTML, CSS, and JavaScript, displaying an interactive live preview directly in your chat. When you like the result, click Save to add it to your dashboard.
For complete end-to-end features, you can describe both the logic and the visual design in one message. The Genie will write the background program and build a widget connected to its output, waiting for real data before showing the finished layout.
If you are working inside the Widget Editor, click Update to apply changes suggested by the AI, or use Save As to create a variation. You can also turn on Auto-update to see AI changes applied live as you chat.
Background automation and scheduling
Gemini is not limited to manual conversations. Your own programs, scripts, and scheduled events can use AI intelligence behind the scenes:
Request.Generate: Generates text (such as daily summaries or custom notifications) for your scripts to use without executing device commands.Request.Execute: Carries out multi-step tasks through HomeGenie's action system, using your chosen permissions and context.Request.StatusandRequest.Cancel: Let calling programs check on an active AI task or cancel it gracefully.
Background requests run quietly without adding clutter to your chat history. They allow your system to process sensor data, manage daily routines, and launch helper tasks automatically.
In the Scheduler, the Genie Command action allows you to trigger AI tasks using everyday language. You can assign the task to Gemini Genie, another provider, or Default (which uses your system's default AI configuration). The scheduler waits for the AI task to finish before moving to the next step, ensuring your routines run in reliable sequence.
Choosing between Cloud AI and Local AI
Both Gemini Genie and Lailama share HomeGenie's underlying architecture, permissions system, and device controls. The difference lies in where the processing happens:
| Feature | Gemini Genie | Lailama |
|---|---|---|
| Where it runs | Google's cloud servers. | Your local HomeGenie computer. |
| Privacy & data | Selected context is sent to Google for processing. | Model inference runs locally, without sending prompts to a cloud AI provider. |
| Hardware requirements | None; HomeGenie only runs its lightweight core software. | Requires sufficient RAM, CPU, and optional GPU power for local models. |
| Internet connection | Required for all AI queries. | Works completely offline once model files are downloaded. |
| Speed & capability | Very fast reasoning, backed by Google's latest frontier models. | Depends directly on your local computer's processing power and model size. |
You do not have to choose just one. You can use Gemini Genie for coding complex programs and designing widgets, while relying on Lailama for local AI tasks without a cloud connection.