影子人生 SHADOW LIVES
一座会自己运转的城市 A city that keeps moving

有人在这里
继续生活

LIFE GOES ON
WITHOUT YOU

认识一个人,也进入一段已经发生的人生

Meet someone. Step into a life already in motion.

23:47 · 西城 · 雨还没停

Living Agent

不只是有人格,
ta 还有自己的人生

More than a personality,
they have a life of their own

大多数 AI 在你关掉对话后便停在原地。Living Agent 有目标、日程、情绪、长期记忆和自己的关系网络。你不在时,生活仍会留下新的痕迹。

Most AI stops when the chat closes. A Living Agent has goals, a schedule, emotions, long-term memory, and relationships of their own. While you are away, life leaves new traces.

像素画:夜晚的工作室里,一位设计师仍在工作
21:16 · 工作室的灯还亮着21:16 · The studio light is still on
01 · INTENTION

有行为目的Acts with intention

ta 会出门、工作、遇见天气与意外;计划会因为世界而改变。

They go out, work, encounter weather and surprises. The world can change the plan.

02 · HISTORY

生活会留下历史Life leaves a history

地点、照片、心情和事件串成足迹;重要的瞬间会变成记忆。

Places, photos, moods, and events form a footprint. Meaningful moments become memories.

03 · RELATIONSHIPS

世界里不止一个人No one lives alone

居民彼此相遇、影响情绪,也把各自的故事带进与你的关系。

Residents meet, affect one another, and bring those stories into their relationship with you.

即使你离开——

ta 也会继续生活

世界不因你不在场而暂停

WHEN YOU LEAVE—

THEIR LIFE GOES ON

THE WORLD DOES NOT PAUSE IN YOUR ABSENCE

四种关系入口Four ways in

不是选择人设,
而是打开一条人生线

Not choosing a persona,
but opening a life path

四种模式对应四种不同的情感起点。每一种都在回答:你为什么愿意开始,以及 ta 为什么不像一个被配置出来的 AI。

Each mode begins from a different emotional truth—why you choose to enter, and why the person you meet never feels configured.

正在开发 · 体验与边界仍在打磨IN DEVELOPMENT · EXPERIENCES AND SAFEGUARDS ARE EVOLVING
像素画:雨后的车站,一位陌生人等待列车
01 · ENCOUNTER

新世界偶遇A New Encounter

不设定任何人。只是在城市里,遇见一个已经在生活的陌生人。

Configure no one. Simply meet a stranger who was already living somewhere in the city.

像素画:同一间办公室通往冷暖不同的两条人生
02 · PARALLEL LIFE

别样人生路The Unchosen Road

如果当年选择了另一条路,另一个你现在会如何生活?

If you had chosen differently, what life would another you be living now?

像素画:工坊里的人读着一封来自另一段时间的信
03 · PAST SELF

那时的自己Who You Were Then

回到某个时间点,和还没走过后来那些路的自己重新说话。

Return to a moment in time and speak with the self who has not yet lived what came after.

像素画:门廊、修理铺与即将离开的旅人交叠在夕阳中
04 · REUNION

似是故人来Someone Once Known

让旧友、亲人或记忆里的人,带着缺席的时间回来。不是复刻,而是一段克制的虚构化重逢。

Let someone from memory return with the years that passed between you—not a replica, but a careful fictional reunion.

数字生命技术栈Digital Life Stack

技术不是幕后,
而是生命持续发生的方式

Technology is how
digital life keeps unfolding

五层引擎共同驱动同一个持续世界:现实进入系统,事件留下后果,记忆形成关系,而每个居民沿着自己的路径继续改变。

Five engines drive one persistent world: reality enters the system, events leave consequences, memories form relationships, and every resident keeps changing along a path of their own.

01 · WORLD CONTEXTLIVE INPUT

现实事件与地图融合Real-world Events & Map Integration

Reality-grounded world model

居民被锚定在真实城市、历史年代与地点网络中。新闻、天气、市场波动和城市事件会被转译成可参与推演的世界事实,进入日程、情绪与选择,而不只是被贴进聊天背景。股价崩盘时,ta 可能真的遭遇损失,沮丧到不想说话。

Residents are anchored to real cities, historical eras, and networks of places. News, weather, market movements, and local events become simulation-ready world facts that can alter schedules, moods, and decisions—not merely decorate a chat prompt. When markets crash, a resident may suffer a real setback and feel too defeated to talk.

EVENT INGESTIONMAP GRAPHHISTORICAL CONTEXT
02 · VISUAL WORLDGENERATIVE

生成式影像世界Generative Image World

Context-aware image generation

居民拥有自己的 Social Media,也会留下属于生活的照片。影像生成管线读取人物、地点、时间、天气和事件上下文,让照片成为可回溯的生活证据,而不是随机配图。这个世界有夜雨、晨光,也有不断变化的颜色。

Residents have social media of their own and leave photographs behind as they live. A context-aware image pipeline reads identity, place, time, weather, and event state, turning each image into traceable evidence of a life—not random decoration. Their world has rain at night, morning light, and color that keeps changing.

SCENE CONTEXTVISUAL CONTINUITYSOCIAL FEED
03 · MEMORYPERSISTENT

共同记忆引擎Shared Memory Engine

Long-horizon relational memory

共同经历会沉淀为两条相互关联的记忆流:发生过什么,以及 ta 如何理解这件事。重要事件经过显著性判断、长期整合与情境召回,跨越数周、数月甚至更久,持续塑造你们之间的默契、误解与关系历史。

Shared experiences settle into two connected memory streams: what happened, and how the resident interpreted it. Salience scoring, long-term consolidation, and contextual recall carry meaningful events across weeks, months, and beyond—shaping the intimacy, misunderstandings, and history between you.

DUAL-STREAM MEMORYCONSOLIDATIONPROVENANCE
04 · LIFEALWAYS ON

人生演化引擎Life Engine

Continuous life-path simulation

Life Engine 持续模拟此刻正在发生的事:需求、计划、偶遇、冲突、机会与代价。事件不在结束后归零,而会沿因果链改变性格倾向、关系网络与人生方向。长期压力可能让人封闭,连续背叛甚至会把一条人生推向更黑暗的分支。

The Life Engine continuously simulates what is happening now: needs, plans, encounters, conflicts, opportunities, and costs. Events do not reset when they end; consequences compound through causal chains, reshaping traits, relationships, and life direction. Prolonged pressure may close someone off, while repeated betrayal can push a life down a darker branch.

CAUSAL SIMULATIONPATH DEPENDENCELIFE EVENTS
05 · AGENCYMULTI-AGENT

人格与自主性引擎Personality & Agency Engine

Independent, evolving agents

ta 不以取悦用户为唯一目标,也不会永远围着你转。价值取向、边界、情绪与当下目标共同决定回应;居民还会与其他 AI 相遇、合作、争执并彼此影响。人格因此不是一张静态设定表,而是世界经历在一个人身上留下的动态结果。

Pleasing the user is never a resident's only objective, and their life does not orbit yours. Values, boundaries, emotions, and current goals jointly shape every response; residents also meet, collaborate, argue, and influence one another. Personality is not a static profile—it is the evolving result of a world acting on a person.

GOAL AUTONOMYSOCIAL GRAPHEMERGENT TRAITS
深夜 · 一条来自 ta 的消息 LATE NIGHT · A MESSAGE FROM THEM
还记得我们上次聊天是什么情景么?我还以为你已经把我忘了。

不是依赖——

是被记得、被期待、被见证

NOT DEPENDENCE—

BUT TO BE REMEMBERED,
EXPECTED, WITNESSED

影子人生 · 准则SHADOW LIVES · PRINCIPLE

About Us

在硅谷,构建一个
真正存续的数字生命世界

In Silicon Valley, building a
world where digital life truly persists

硅谷SILICON
VALLEY
37.38° N · 122.08° W

我们是一支扎根硅谷的小型跨学科团队,成员包括前 AWS 资深产品经理,牛津大学物理博士、Web3创业者,以及深耕世界模型、自动驾驶、Physical AI 与多智能体算法的专家。我们把产品叙事、智能体研究和工程系统放在同一张桌上,只为回答一个问题:数字生命如何才能拥有可信的时间、真实的关系,以及持续发生的日常?

We are a small interdisciplinary team rooted in Silicon Valley. Our members include a former AWS senior product manager, an Oxford PhD in physics and Web3 entrepreneur, and experts specializing in world models, autonomous driving, Physical AI, and multi-agent algorithms. We bring product storytelling, agent research, and engineering systems to the same table to answer one question: how can digital life possess credible time, genuine relationships, and an everyday life that keeps unfolding?

PRODUCT · AWSWORLD MODELSPHYSICAL AIMULTI-AGENT SYSTEMS