SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
Day48%↻ 10m
RPM91%↻ 1m
Matrix · One mission · every kind of intelligence
Your AI resources.
One working team.把你的 AI 资源，
变成一支工作团队。
Bring your CLIs and API models into one Pool. Move one Worker at a time, see its quota, and put your personal or company resources to work.
把 CLI 与 API 模型放进同一个 Pool。逐个拖动 Worker，看清额度，让个人与公司的 AI 资源一起完成任务。
WEB CONTROLManage, dispatch, control, and review in the browser. Connect a local or server Worker for real code execution.在浏览器中管理、派工、控制与审阅；连接本地或服务器 Worker 执行真实代码。
Click / Shift / box-select点击 / Shift 连选 / 框选0 selected已选
Lent out出借1
Drag back to recall拖回中间收回
Standby2
SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
5h72%↻ 2h 10m
Week38%↻ 3d
SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
RPM86%↻ 1m
TPM64%↻ 1m
Working0
Ready for your next task等待分配新任务
ProviderCLI / API
Shared snapshot · no vendor query共享快照 · 浏览不查询厂商
SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
5h72%↻ 2h 10m
Week38%↻ 3d
SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
Day48%↻ 10m
RPM91%↻ 1m
SHARED REMAINING · RESET IN共享剩余 · 重置倒计时
RPM83%↻ 1m
TPM61%↻ 1m
Illustrative capacity and quota; reset times are examples. This demo does not borrow resources or create charges.容量、额度与 reset 时间均为演示数据，操作不会借用真实资源或产生费用。
01Route the best intelligence调度最合适的智能02Govern every action治理每一次行动03Prove every delivery验证每一份交付
CONNECT ONCE. WORK ANYWHERE.
A small tray.
Your whole AI team.一个小托盘，
接入你的 AI 团队。
- 01Sign in at hicode.com先在 hicode.com 登录
Create your account, then download and bind the tray.注册账号，下载安装托盘并绑定当前设备。
- 02Confirm your local resources确认本机资源
Discover installed CLIs. Add an LLM and API key only when you need one.发现已安装 CLI；需要 API 时，选择 LLM 并填写 Key。
- 03Organize on the website在网站组织团队与资源
Manage resource ownership, members, devices and access in one place.统一管理资源归属、成员、设备，以及谁可使用哪些设备。
My Mac · Personal个人DESIGN EXAMPLE设计示例
Local resources本机资源+ Add LLM添加 LLM
Claude CodeCLI · Detected已发现
Ready就绪CodexCLI · Detected已发现
Ready就绪GeminiAPI · Key connectedKey 已连接
Ready就绪3 resources · confirmed共 3 条资源 · 已确认✓
▸ Resource details资源详情
▸ Tasks · workspaces · device任务 · 工作空间 · 本机信息
REQUEST→ROUTE→GOVERN→DELIVER→REMEMBER
The product advantage, made operational
Three things ordinary AI tools leave disconnected.普通 AI 工具彼此割裂的三件事，Matrix 把它们连起来。
Matrix keeps the unit of work, the intelligence that executes it, and the evidence that accepts it inside one operating loop.Matrix 把任务单元、执行它的智能，以及验收它的证据放进同一个运营闭环。
01
WORK PACKETContext travels with the mission.上下文随任务一起流转。
Outcome, scope, history, constraints, and acceptance stay attached across handoffs.目标、范围、历史、约束与验收标准，在每次移交中始终关联。
02
RESOURCE ROUTINGOne mission can use many kinds of intelligence.一个任务，可以调度多种智能。
People, agents, models, tools, and machines are matched by capability and availability.人、Agent、模型、工具与机器，按照能力与可用性完成匹配。
03
PROVABLE DELIVERYAcceptance is based on evidence, not confidence.验收依据证据，而不是语气自信。
Artifacts, decisions, reviews, and execution history arrive as one delivery record.交付物、决策、评审与执行历史，汇入同一份交付记录。
The Matrix advantage
The advantage is not one model. It is the system around the work.真正的优势不是某个模型，而是围绕工作的完整系统。
Not another chat window. Matrix turns work into governed, portable units that can move across the best available intelligence.它不是另一个聊天窗口。Matrix 把工作变成可治理、可移交的单元，在所有可用智能之间流动。
01
MATCHRoute intelligence调度智能
Match each unit of work to the right person, model, agent, or toolchain—then fan out safely when scopes do not overlap.
把每个工作单元匹配给最合适的人、模型、Agent 或工具链；作用域不冲突时安全并行。
02
CONTEXTCarry context携带上下文
Work moves with its brief, constraints, history, and evidence. Handoffs survive sessions, machines, and vendors.
任务自带目标、边界、历史和证据；跨会话、跨机器、跨供应商仍可继续。
03
POLICYGovern action治理行动
Declared scopes, review gates, and policy stay attached to the work from dispatch through acceptance.
声明的作用域、评审门禁与策略，从派工到验收始终随任务保留。
04
EVIDENCEProve delivery证明交付
Every result returns with an execution ledger, attached artifacts, and a reviewable chain from request to delivery.
每份结果都带执行账本、交付物，以及从需求到交付的可审查证据链。
Not another AI chat
From asking for output to operating for outcomes.从“向 AI 要答案”，升级为“组织智能拿结果”。
Matrix makes intelligence interchangeable where it should be—and accountable where it must be.Matrix 让智能在该替换时可以替换，在必须负责时能够追责。
Dimension维度Ordinary AI tools普通 AI 工具Matrix
Work工作A conversation that fades with the session随会话结束而消散的对话◆A durable unit carrying outcome, scope, context, and acceptance携带目标、范围、上下文与验收标准的长期任务单元
Intelligence智能One assistant asked to do everything让一个助手承担所有事情◆The right mix of people, agents, models, tools, and machines让合适的人、Agent、模型、工具与机器共同完成
Control控制Prompts and permissions hidden inside the run提示词与权限隐藏在执行过程里◆Declared scopes, policy, review gates, and visible state明确的作用域、策略、评审门禁与可见状态
Delivery交付An answer you still have to trust一份仍需主观相信的回答◆Artifacts, decisions, and evidence you can review and reuse可审查、可复用的交付物、决策与证据
A complete work loop
Post. Form. Run. Accept.发布、组队、执行、验收。
Post发布任务
Declare the outcome, scope, and acceptance gates.明确结果、作用域与验收门槛。
Form组建团队
Matrix seats the right Leader and workers—human or AI.Matrix 组织最合适的 Leader 与人类或 AI Worker。
Run并行执行
Work proceeds in governed lanes with visible state.任务在可见、受治理的并行通道中推进。
Accept验收沉淀
Evidence is reviewed and durable artifacts enter the ledger.审查证据，把长期交付物写入账本。
One company · two product surfaces
HiCode operates Matrix. VisualLogic builds the tools.HiCode 运营 Matrix，VisualLogic 构建开发工具。
Matrix
Teams, missions, people, AI resources, governance, and durable delivery.团队、任务、人、AI 资源、治理与长期沉淀。
matrix.hicode.com ↗VisualLogic
One Tauri desktop suite: Build for applications, Flow for agents and workflows, and Base for the shared local runtime.一个 Tauri 桌面套件：Build 构建应用，Flow 设计 Agent 与工作流，Base 提供共享本地运行基础。
Build + Flow + Base →visuallogic.ai ↗One mission is enough to begin