Kickstarter story

Lumi AI Dock: The First Desktop AI That Physically Uses Your Phone

Your AI agent can browse the web. Now it can use your phone. Real phone. Real screen. Real taps.

Live phone sessionPermission on
tap
AI memory outputTodo created

Send investor update by Friday

Source: Phone chat 09:41
observe_screen()tap target confirmedwrite Todo + trace

1

Your Phone Is Where Work Actually Happens

Important promises, links, files, screenshots, private chats, booking steps, and relationship context are trapped inside mobile apps. Lumi gives AI a supervised physical bridge into that reality.

2

Meet Lumi: See, Tap, Remember

See the real phone screenTap, swipe, and type physicallyRemember Todo, People, Notes, Calendar

3

Demo: Messages Become Todo

Play
Messages -> Todo

Phone chat, physical tap, source-traced task.

4

Demo: People Memory

Play
People memory

Waiting reply, promised quote, next meeting.

5

Demo: Notes and Calendar Memory

Play
Notes and Calendar

Screenshots, files, links, and reminders organized.

6

Real Phone. Real Screen. Real Taps.

Vision module

Sees the real phone screen

4-axis mechanism

Positions for real taps and swipes

Touch actuator

Physical interaction with the screen

Phone dock

Stable alignment for common phones

AI head

Session state, voice, and local controls

7

PC Phone Copilot

Confirmation requiredSend this message?

Lumi asks before sensitive or irreversible actions.

09:41 observed Messages09:42 opened source thread09:42 created Todo draft09:43 waiting for approval

8

Local-First Privacy

Local-first by defaultNo root or jailbreakVisible action logUser confirmationNo app hacking

Lumi does not hack apps. It sees and taps like a human, with your permission.

9

Phone Use SDK

session = phone_use.connect("lumi-dock")
screen = session.observe_screen()
target = screen.find("Confirm appointment")

session.confirm_action("Tap confirmation button?")
session.tap(target)
session.type("Friday works. Sending calendar invite.")

memory.write({
  type: "todo",
  source: session.screenshot(),
  trace: session.action_log()
})

10

Interactive AI Head and Voice

Lumi's head is a serious status surface for session state, voice input, permission prompts, and visible feedback. It is not designed as a toy companion.

11

Hardware and Specs

Phone accessReal screen observation with physical tap, swipe, and type
Mechanism4-axis physical control concept
PrivacyLocal-first memory and user-confirmed actions
MemoryTodo, People, Calendar, Notes
SDKobserve_screen(), tap(), swipe(), type(), screenshot(), confirm_action()
CompatibilityValidation in progress for common iOS and Android sizes

12

Rewards

Reward

Early Bird Lumi AI Dock

Best price for early Kickstarter backers.

Dock, Lumi app, core memory workflows
Reward

Lumi AI Dock Pro / Creator Kit

For demo builders, creators, and power users.

Pro accessories, creator workflow templates
Reward

Developer Kit

Early hardware-level phone access for agent builders.

SDK preview, examples, developer updates
Reward

Duo / Team Pack

For founders, sales teams, and small studios.

Two docks, team onboarding, shared playbooks

13

Timeline

2026 Q2

Working prototype demo

2026 Q3

EVT hardware validation

2026 Q4

DVT + Lumi App Beta

2027 Q1

PVT and pilot production

2027 Q2

Mass production

2027 Q2/Q3

Shipping begins

14

Risks and Limitations

Will do at launch

Real phone observation

Basic tap, swipe, type

User supervision and confirmation

Todo / People / Notes / Calendar basics

Basic Phone Use SDK

Won't promise

Full automation for every app

Unattended sending, payment, or deletion

Every phone model

Fully local large model for all tasks

Enterprise-grade RPA stability

App hacking or encryption bypassing

15

Back Lumi on Kickstarter

Follow now for launch notification, early bird rewards, and prototype progress.