LUMI 1 Pro
An autonomous presence at home.
It sees you walk in — and nothing about you ever leaves. The head turns to meet you, and it speaks first.
Pre-order — ships 2027
$120.00
Specs at a glance
- Face
- 1.75″ AMOLED — emotion, expressions, quiet hours dimming
- Vision
- 2 MP camera, processed entirely on-device
- Motion
- Pan 180° + tilt · two brushless motors
- Audio
- Two MEMS mics · full-range 40mm speaker
- Connectivity
- Wi-Fi 6 + BLE 5.3
- Power
- USB-C, 5V/3A
In the box
- LUMI 1 Pro
- In Midnight Black or Lunar White
- USB-C cable
- One cable, any 5V/3A adapter
- Quick-Start card
- Your unique pairing QR
Cat Ears and other cosplays ship separately.
Delivery & pre-order
Free shipping during the pre-order window · ships 2027. Nothing is charged until our payment provider captures payment before shipping; tax and duties are calculated at checkout.
The tech
Explore
LUMI 1 Pro
On-device vision, a motorised head, and a memory that knows what changed — a presence, not a camera.
It sees you. On device.
A camera and recognition that run entirely on the robot's own silicon: it turns to face whoever speaks, tracks you across the room, and glances away while it thinks. No frame is uploaded, stored, or clipped — the useful part of vision stays, the surveillance part was designed out.
Tunable presence
Two brushless motors with magnetic encoders give the head 180° of pan and a tilt axis — it meets your eyes and follows the speaker. On power loss it simply holds its position and saves its state. Pet it and it leans in; pick it up and it knows.
The camera that never uploads
No telepresence, no live feed, no cloud activity clips — those were designed out, not switched off. Audio is ephemeral by construction (stream, transcribe, discard the raw buffer), and the double-tap mute drops that buffer before the network stack sees it. The red LED is the guarantee.
“It is not a tool you query; it is an entity that shares your physical space.”
— AMOLUMI · the north star
A day it notices
The room's quietest participant.
- 08:40
- You sit down at your desk and it already noticed: the head turns to meet you, the lamp comes up to Focus, your playlist queues, and the greeting is in the voice you chose — with yesterday's late meeting mentioned exactly once.
- 15:10
- It is talking when you start talking. Under fifty milliseconds it stops mid-sentence, listens, and answers — then glances away while it thinks. That glance is the whole trick: it behaves like a presence, not a camera.
- 21:50
- The room empties and it dozes. Lights to your evening scene, a summary of tomorrow set in memory, and the face settling to sleepy as quiet hours begin.
Presence triggers, timed routines, and your own voice macros ("code time" sets the whole room in one sentence) run whether or not anyone is watching — because nothing is.
What it does
- Vision
- A camera and recognition that run entirely on the robot's own silicon — no frame is ever uploaded, stored, or clipped
- Motion
- A motorized head: pan 180° with tilt, two brushless motors and magnetic encoders — it turns to whoever speaks and follows you with its face
- Face
- 1.75″ AMOLED — smiles when you arrive, glances away when thinking, dozes when alone
- Personality
- Realtime conversation with barge-in under 50ms — it hears you while it speaks, and knows a pause from a finished thought
- Audio
- Two MEMS mics and a full-range 40mm speaker tuned per personality; double-tap mutes the mic and lights a solid red LED
- Home control
- Lights and scenes across four local tiers — WiZ, LIFX and Govee over your LAN, Tapo, Hue, Aqara and Nanoleaf over Matter, legacy Tuya, and a Home Assistant bridge
- Offline
- Wake word and 200 voice commands run on the device itself — timers, alarms, and every home-control tier keep working when the internet does not
- Memory
- Persistent, per-speaker memory with validity windows — it recalls what changed, not just what was said
- Touch
- Three capacitive pads — pet it and it leans into your hand and purrs; double-tap for mute; hold and it knows it has been picked up
- Skins
- Magnetic snap-on cosplays (8 neodymium points, 5-second swap) that bring their own face, voice, sounds, and speaker EQ
- Power loss
- The head holds its position and the robot saves its state — nothing is lost when the lights go out
Good to know
Questions, answered.
Where does the camera video go?
Nowhere. Vision is processed on the robot itself — no frames are uploaded, stored, or clipped. There is no remote camera feed and no cloud activity reel, by design rather than by setting.
Does it move around the house?
No — and we say so plainly. LUMI 1 Pro stands still and moves its head: 180° of pan and a tilt axis, so it turns to face you and follows you with its face. It does not drive around the floor.
Can it hear me while it talks?
Yes — barge-in under 50ms. It stops mid-sentence when you start yours, and it knows a pause from a finished thought. Double-tap the head and the mic is off: solid red LED, audio buffer dropped before the network stack.
What colors does it come in?
Midnight Black and Lunar White — same price, same everything else. Pick yours above; you can swap the colourway in your cart before checkout.
What happens when the internet drops?
The presence stays. Wake word, 200 offline commands, alarms, timers, and all four home-control tiers run on the device. Only conversation waits for the connection to return.
Will my cat ears fit both models?
Yes. Cosplay attachments are universal across the LUMI 1 family — snap on, swap in seconds, and the whole robot changes personality with them.
When does it ship?
Pre-orders ship in 2027. Nothing is charged until our payment provider captures it before shipping.
Bring one home.
Pre-order LUMI 1 Pro in Midnight Black or Lunar White — ships 2027.