Recent Home Assistant searches show growing interest in AI devices and AI automation, but a useful smart-home result still depends on accurate room context. A language request such as “make the office comfortable when I am working” needs a presence entity, a climate target, a time rule and a clear boundary around what can change.
This guide explains a room-first workflow for LinknLink App Agent+ and supported Home Assistant setups. It connects AI Agent planning with mmWave presence, temperature context and IR appliance control, while keeping the proposed actions visible before they are approved.

What a Room-First AI Agent Scene Can Coordinate
Room context
Choose the area, presence zone, climate state and time window that define the scene.
Action plan
Separate lighting, climate, IR and notification actions so each one can be reviewed.
Human approval
Test a small reversible step before enabling a multi-device room routine.
A room scene is easier to understand when it has one location, one occupancy signal and a short list of device actions. The AI Agent can help turn that goal into a proposed plan, while Home Assistant supplies the entities, states and automation conditions for review.
Before You Start
- Create a clear Home Assistant area and use room-first entity names for sensors and appliances.
- Confirm which presence, temperature, humidity and climate entities are available to the integration.
- Decide whether the first scene should notify, change one light, adjust climate or send one IR command.
- Define occupied, vacant, unavailable and manual-override behavior before asking the Agent to prepare a plan.
- Keep locks, alarms, cameras, purchases, messages and other sensitive actions outside the first scene.
- Check the current LinknLink App, Home Assistant integration and device permissions before testing.
Build, Preview and Test One Room Scene
1. Name the room outcome
Write the desired result in plain language, such as keeping the office light and temperature comfortable while desk presence remains active.
2. Map the context entities
Select the room presence sensor, climate entity, temperature or humidity source, time helper and any manual override helper.
3. Split the proposed actions
Ask the Agent to list lighting, climate, IR and notification actions separately rather than hiding them in one broad scene.
4. Review conditions
Check the occupancy threshold, vacancy delay, time window, current mode and what should happen when an entity is unavailable.
5. Preview the IR command
If the scene uses an air conditioner, TV or fan, verify the selected appliance, command and room before approval.
6. Run a low-impact test
Start with a notification or one light action, then observe the state transition while entering, sitting in and leaving the room.
7. Approve the expanded scene
Only after the test behaves as expected should you enable climate or IR actions, with a manual override and rollback path.
Example Prompts and What to Review
| Use case | Example request | Review before action |
|---|---|---|
| Office focus scene | When office-presence is occupied during work hours, prepare the desk light and climate plan, then show both target entities before approval. | Room, time window, presence state and two proposed actions |
| Living room comfort | When living-room-presence remains occupied for 10 minutes, suggest a temperature adjustment without changing the setpoint until I approve it. | Delay, climate target, current mode and approval |
| IR air-conditioner scene | Use the living-room presence and temperature state to prepare one IR cooling command, and show the appliance and command before running it. | IR target, command, room and fallback |
| Night bedroom scene | After 10 PM, prepare a gentle bedroom light scene when presence is detected, but do not change the climate or access devices. | Time rule, excluded devices and manual override |
| Away review | When all monitored rooms are vacant, summarize possible climate and light changes without switching anything off automatically. | Read-only summary, room list and no automatic action |
| Unavailable sensor | Show the room scene plan when the presence sensor is unavailable, and do not treat missing data as occupied or vacant. | Fallback state, notification and safe default |
Use Read, Prepare, Approve and Execute Stages
| Stage | Agent task | User control |
|---|---|---|
| Read | Inspect only the permitted source. | Choose the app, date range, room, folder or account. |
| Prepare | Extract entities and build a draft or proposed plan. | Check assumptions, missing information and affected resources. |
| Approve | Present the action and destination. | Confirm or revise before execution. |
| Execute | Run the supported action after approval. | Verify the result and keep a reversal path when possible. |
Review Checklist
- Does the scene describe one room and one expected outcome?
- Are the presence, temperature, humidity, climate and helper entities mapped to the intended room?
- Are lighting, climate, IR and notification actions listed separately?
- What happens when a sensor is unavailable or a person remains still?
- Is the vacancy delay long enough for the way the room is used?
- Does the IR target and command match the intended appliance?
- Can a person override the scene and undo the action quickly?
- Did the low-impact test work before climate or IR actions were approved?
Privacy, Permissions and Safer Defaults
Cross-app tasks can touch personal communications, calendars, files, photos, locations and household devices. Grant the smallest useful permission, keep unrelated data out of scope and avoid sharing credentials, passwords, one-time codes or private account details.
Use drafts for messages and attachments. Use scene previews for smart-home changes. Keep purchases, access control, security devices and destructive file actions behind explicit human confirmation.
Troubleshooting
The scene uses the wrong room
Add the exact Home Assistant area and entity names, then ask the Agent to show its selected entity list before execution.
The light turns off while someone is seated
Review the presence zone and vacancy delay, test the sensor while remaining still, and separate the low-impact light test from climate actions.
Climate changes too often
Add a minimum hold time, temperature deadband or helper condition, and ask the Agent to show the repeat-trigger behavior.
The IR command is not the intended one
Confirm the appliance, room and command in the preview, then test one manual IR command before adding it to a presence scene.
The sensor is unavailable
Use an explicit unavailable path, avoid converting missing data into occupied or vacant automatically, and check the integration connection.
The Agent proposes too many actions
Split the scene into read, prepare, approve and execute stages, starting with one room and one reversible action.
Related LinknLink Products
- HomeClaw: a local Home Assistant and AI gateway path.
- eMotion Ultra: 60GHz presence, temperature and humidity sensor cable support, and built-in IR.
- eMotion Air: battery-powered mmWave presence context for room automations.
- eMotion Pro: mmWave presence sensing and supported room automation signals.
- eRemote HA: local IR control for compatible appliances in Home Assistant.
Related LinknLink App and Home Assistant Guides
- Home Assistant AI Agent safety checklist
- LinknLink Agent+ goal-driven workflow guide
- LinknLink App Agent+ diagnostics guide
- HomeClaw local AI voice control guide
- Home Assistant presence sensor calibration guide
- Home Assistant heatwave automation guide
Frequently Asked Questions
Can an AI Agent build a Home Assistant room scene?
A supported AI Agent can help interpret a room goal and prepare a proposed scene. Review the entities, conditions, actions and fallback behavior before enabling it.
Why combine an AI Agent with a presence sensor?
Presence provides room context that helps an Agent or Home Assistant distinguish an occupied room from a simple motion event or an empty room.
Can a room scene control an air conditioner with IR?
A supported Home Assistant and IR workflow can prepare or send an appliance command. Confirm the appliance, command, room and approval state before running it.
Should climate actions run immediately when presence is detected?
Usually review the time window, temperature difference, minimum hold time and vacancy delay first. A preview or suggestion is a useful first test.
How should I handle a missing presence sensor?
Define an unavailable-state path and avoid assuming that missing data means occupied or vacant. Notify the user or hold the scene until the entity returns.
Can eMotion Ultra provide more than presence context?
eMotion Ultra combines 60GHz presence sensing with a Temperature and Humidity Sensor Cable and built-in IR. Verify the current device integration and available entities in your setup.
Start with One Room and One Reversible Scene
Choose one room, map its presence and climate entities, ask the Agent to preview the scene, approve one low-impact action, and verify the state transition before expanding to IR or whole-home routines.