Sentra Smartboard Semayot

Operational command center for Semayot owners, staff, outlets, data, and AI

Sentra Smartboard is the operational decision center for Semayot. It turns sales, inventory, cashier activity, customer signals, reports, and AI summaries into clear daily priorities that owners and staff can act on immediately.

Executive Summary

Sentra Smartboard Semayot is the command-center layer for Rumah Makan Semayot. It is not positioned as a simple restaurant website. It is an operational work system that connects four surfaces inside one codebase:

Owner/staff Smartboard at /admin/overview
Admin Workspace for menu, POS, transactions, reports, customers, SEO, and settings
AI + Intelligence Layer for business summaries, owner consultation, memory, priorities, and public chat
Public Restaurant Site as the brand entry point and customer conversation surface

This README intentionally focuses on Sentra Smartboard, because that is where Semayot's daily decisions are formed: operational numbers are read, risks are surfaced, priorities are selected, AI explains context, and the owner or staff moves into the right action surface.

Product Thesis

A good dashboard does not merely show numbers. A correct dashboard helps the owner take the right action on the same day.

Sentra Smartboard is built to answer four operational questions:

Owner Question	Smartboard Answer
How is the outlet performing today?	`Nadi Outlet`, KPIs, transactions, cashier closing, and data status
What needs to be handled first?	`Today's Priorities`, `Stock Risk Radar`, and deterministic recommendations
Which risks must not be missed?	critical stock, unsafe closing, weak transactions, missing data
How can AI help without inventing facts?	SEMA Daily Brief, quick prompts, owner chat, public chat, memory inbox

Sentra Design Language

The README and product direction follow the Sentra Smartboard language: warm, tactical, operational, and precise.

Element	Sentra Direction
Visual tone	Warm operational command center
Background	Blush-pink, warm-neutral, and high-contrast cards
Accent	Coral, rose, amber, violet, sky, emerald
Typography	Strong display headlines, mono labels for status and signals
Component shape	Flat boxes, tactical chips, command-center cards
Motion	Used for orientation and emphasis, not empty decoration
AI posture	Helps the owner think; does not replace business judgment

Executive Summary
Product Thesis
Sentra Design Language
System Map
Smartboard Command Loop
Product Surfaces
Smartboard Modules
AI and Intelligence Boundary
Data Truth Model
Role and Access Boundary
Route Map
Architecture Folder Map
Local Runbook
Environment Variables
Testing and Governance
Most Important Files
Product Truth Principles

System Map

flowchart LR
    Public["Public Site<br/>Homepage + Visitor Chat"]
    Smartboard["Sentra Smartboard<br/>/admin/overview"]
    Workspace["Admin Workspace<br/>Menu, POS, Transactions, Reports"]
    APIs["App Router APIs<br/>/app/api/**"]
    Supabase["Supabase<br/>Auth + Tables + RLS"]
    Deterministic["Deterministic Intelligence<br/>Scoring + Priority Engine"]
    GenAI["Generative AI Layer<br/>Gemini via AI SDK"]
    Owner["Owner / Staff<br/>Operational Decisions"]

    Public --> APIs
    Smartboard --> APIs
    Workspace --> APIs
    APIs --> Supabase
    APIs --> Deterministic
    APIs --> GenAI
    Supabase --> APIs
    Deterministic --> Smartboard
    GenAI --> Smartboard
    GenAI --> Public
    Smartboard --> Owner
    Workspace --> Owner

    classDef public fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef board fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:3px;
    classDef workspace fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;
    classDef api fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef data fill:#ECFDF5,stroke:#10B981,color:#064E3B,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef user fill:#F8FAFC,stroke:#475569,color:#0F172A,stroke-width:2px;

    class Public public;
    class Smartboard board;
    class Workspace workspace;
    class APIs api;
    class Supabase data;
    class Deterministic data;
    class GenAI ai;
    class Owner user;

Smartboard Command Loop

Sentra Smartboard is not a static dashboard. It is a decision loop.

flowchart TD
    A["Raw Outlet Signals<br/>sales, inventory, cashier, customer, expense"] --> B["Server Summary Route<br/>/api/admin/overview/summary"]
    B --> C["DashboardOverviewData<br/>raw operational contract"]
    C --> D["Outlet Adapter<br/>normalization + mapping"]
    D --> E["OutletDashboardData<br/>Smartboard contract"]
    E --> F["Nadi Outlet<br/>health score 0-100"]
    E --> G["Today's Priorities<br/>daily action queue"]
    E --> H["Stock Risk Radar<br/>critical, warning, safe, unknown"]
    E --> I["Cashier Closing<br/>cashier close status"]
    E --> J["SEMA Daily Brief<br/>operational summary"]
    J --> K["Quick Prompts<br/>owner asks SEMA"]
    K --> L["Owner Chat Panel<br/>streaming explanation"]
    L --> M["Action Surface<br/>inventory, POS, reports, AI"]

    classDef raw fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef route fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef contract fill:#F5F3FF,stroke:#7C3AED,color:#312E81,stroke-width:2px;
    classDef module fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:2px;
    classDef action fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;

    class A raw;
    class B route;
    class C,D,E contract;
    class F,G,H,I,J module;
    class K,L ai;
    class M action;

Product Surfaces

mindmap
  root((Sentra Smartboard Semayot))
    Admin Overview
      Nadi Outlet
      KPI Summary
      Today's Priorities
      Stock Risk Radar
      SEMA Daily Brief
      Cashier Closing
      Overview Chat
      Analytics
      Transactions
    Admin Workspace
      Menu
      POS
      Transactions
      Reports
      Customers
      Customer Trends
      AI View
      SEO
      Settings
    Intelligence
      Deterministic Scoring
      Stock Risk
      Daily Priorities
      Empty State
      Data Contract
    AI
      Public Chat
      Owner Summary
      Owner Chat
      Memory Inbox
      Quick Prompts
    Public Site
      Homepage
      Hero Media
      Restaurant Info
      Visitor Chat
      llms.txt

Smartboard Module Topology

flowchart TB
    Page["Overview Page Client<br/>Smartboard Shell"]

    Hero["OverviewHero<br/>operator greeting"]
    KPIs["OverviewKpis<br/>sales, transactions, AOV, satisfaction"]
    Health["OutletHealthCard<br/>Nadi Outlet"]
    Priority["TodayPriorityPanel<br/>daily action queue"]
    Stock["StockRiskRadar<br/>inventory risk"]
    Brief["SemaDailyBrief<br/>SEMA action brief"]
    Closing["CashierClosingAssistant<br/>cashier close state"]
    Chat["OverviewChatPanel<br/>owner/staff consultation"]
    Analytics["OverviewAnalytics<br/>operational charts"]
    Transactions["OverviewTransactions<br/>latest activity"]
    Sidebar["OverviewSidebar<br/>user + outlet context"]
    Badge["DashboardModeBadge<br/>live, demo, empty"]
    Empty["SmartEmptyState<br/>honest fallback"]

    Page --> Hero
    Page --> KPIs
    Page --> Health
    Page --> Priority
    Page --> Stock
    Page --> Brief
    Page --> Closing
    Page --> Chat
    Page --> Analytics
    Page --> Transactions
    Page --> Sidebar
    Page --> Badge
    Page --> Empty

    classDef page fill:#111827,stroke:#FF4F79,color:#FFFFFF,stroke-width:3px;
    classDef primary fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:2px;
    classDef metric fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef action fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef support fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;
    classDef fallback fill:#F8FAFC,stroke:#64748B,color:#334155,stroke-width:2px;

    class Page page;
    class Hero,Health,Priority,Stock,Closing primary;
    class KPIs,Analytics,Transactions metric;
    class Brief,Chat ai;
    class Sidebar,Badge support;
    class Empty fallback;

Smartboard Modules

Module	Function	Data Source	Output	Guardrail
`OverviewHero`	Opens operator and outlet context	session + outlet	greeting, role, mode	does not show business claims
`OverviewKpis`	Summarizes outlet performance	sales + transactions	revenue, orders, AOV, satisfaction signal	numbers must come from summary
`OutletHealthCard`	Scores outlet health	outlet contract	0-100 score + reasons	deterministic only
`TodayPriorityPanel`	Builds today's action queue	health + stock + cashier + activity	operational priority list	does not invent priorities
`StockRiskRadar`	Reads stock risk	inventory	critical/warning/safe/unknown	unknown remains unknown
`CashierClosingAssistant`	Checks cashier closing	cashier/shift data	closing status	never assumes safe without data
`SemaDailyBrief`	Summarizes operational actions	deterministic summary	brief + quick prompts	not a fake business diagnosis
`OverviewChatPanel`	Owner/staff dialogue with SEMA	context + prompt	streaming answer	role-aware, context-bound
`OverviewAnalytics`	Visualizes operations	summary data	chart/graph	visuals must not overstate data
`OverviewTransactions`	Shows recent activity	transactions	latest activity	follows transaction data
`DashboardModeBadge`	Marks data mode	app state	live/demo/empty	mode must be visible
`SmartEmptyState`	Honest fallback	missing data	empty state	no fake metrics

AI and Intelligence Boundary

This separation is a core Semayot principle: not every intelligent behavior should come from an LLM.

flowchart LR
    Data["Outlet Data<br/>transactions, stock, cashier, customer"] --> Deterministic["Deterministic Intelligence"]
    Data --> Aggregate["Business Aggregation"]
    Aggregate --> PromptContext["Prompt Context"]
    PromptContext --> LLM["Gemini via AI SDK"]

    Deterministic --> Score["Nadi Outlet Score"]
    Deterministic --> Priority["Daily Priorities"]
    Deterministic --> Stock["Stock Risk Radar"]
    Deterministic --> Empty["Honest Empty State"]

    LLM --> OwnerSummary["Owner AI Summary"]
    LLM --> OwnerChat["Owner AI Chat"]
    LLM --> PublicChat["Public Customer Chat"]

    Score --> Smartboard["Sentra Smartboard"]
    Priority --> Smartboard
    Stock --> Smartboard
    Empty --> Smartboard
    OwnerSummary --> Smartboard
    OwnerChat --> Smartboard
    PublicChat --> Public["Public Site"]

    classDef source fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef deterministic fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:3px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:3px;
    classDef output fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef board fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:3px;
    classDef public fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;

    class Data source;
    class Deterministic deterministic;
    class Aggregate,PromptContext source;
    class LLM ai;
    class Score,Priority,Stock,Empty,OwnerSummary,OwnerChat,PublicChat output;
    class Smartboard board;
    class Public public;

Deterministic Intelligence

Primary locations:

Responsibilities:

Calculate Nadi Outlet from transactions, inventory, cashier closing, cashier activity, menu movement, and anomaly placeholders
Classify stock risk into critical, warning, safe, or unknown
Build daily operational priorities
Generate SEMA Daily Brief while remaining honest when data is incomplete
Produce empty states that do not invent data

Strengths:

fast
deterministic
testable
not token-dependent
suitable for core dashboard logic

Generative AI Layer

Primary locations:

Responsibilities:

Answer customer questions on the public site
Generate owner business summaries by period
Run owner chat using bounded context
Read and mark memory inbox items
Turn Smartboard quick prompts into operational dialogue

Guardrails:

owner AI is limited to the owner role
public chat must not invent restaurant information
summaries require an API key
cache is preferred before regeneration
deterministic signals must not be disguised as LLM output
LLM output must not replace source numbers

Data Truth Model

flowchart TD
    Real["Real Data<br/>Supabase rows"] --> Summary["Server Summary<br/>validated aggregate"]
    Summary --> Contract["Dashboard Contract<br/>typed and normalized"]
    Contract --> Deterministic["Deterministic Output<br/>score, risk, priority"]
    Contract --> AIContext["AI Context<br/>bounded summary"]
    AIContext --> GenAI["Generative Output<br/>explanation, advice, chat"]

    Deterministic --> UI["Smartboard UI"]
    GenAI --> UI

    Empty["Missing / Insufficient Data"] --> EmptyState["Honest Empty State"]
    EmptyState --> UI

    Fake["Invented Number"] -. blocked .-> UI
    FakeAI["Unbounded AI Claim"] -. blocked .-> UI

    classDef truth fill:#ECFDF5,stroke:#10B981,color:#064E3B,stroke-width:3px;
    classDef contract fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef ui fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:3px;
    classDef empty fill:#F8FAFC,stroke:#64748B,color:#334155,stroke-width:2px;
    classDef blocked fill:#FEE2E2,stroke:#DC2626,color:#7F1D1D,stroke-width:2px,stroke-dasharray:5 5;

    class Real,Summary truth;
    class Contract,Deterministic,AIContext contract;
    class GenAI ai;
    class UI ui;
    class Empty,EmptyState empty;
    class Fake,FakeAI blocked;

Truth Levels

Level	Name	Source	Can Be Displayed?	Notes
L0	Raw data	Supabase tables	only through aggregate	should not directly become UI claims
L1	Server summary	API route	yes	primary operational numbers
L2	Dashboard contract	adapter + type contract	yes	safe shape for UI
L3	Deterministic intelligence	scoring/rules	yes	must be testable
L4	Generative AI	Gemini	yes, with AI label	must not replace data
L5	Missing data	null/empty/unknown	yes	display as empty/unknown
L6	Invented data	no source	no	must be blocked

Role and Access Boundary

flowchart LR
    Visitor["Visitor"] --> Public["Public Homepage"]
    Visitor --> PublicChat["Public Chat"]

    Staff["Staff"] --> StaffOverview["Limited Overview"]
    Staff --> POS["POS"]
    Staff --> Transactions["Transactions"]

    Owner["Owner"] --> Smartboard["Full Sentra Smartboard"]
    Owner --> Reports["Reports"]
    Owner --> AI["Owner AI"]
    Owner --> Branch["Branch Switching"]
    Owner --> Settings["Settings"]

    Middleware["Middleware<br/>/admin/* + /api/admin/*"] --> Session["Session Check"]
    Session --> RoleCheck["Route-level Role Check"]

    RoleCheck --> Staff
    RoleCheck --> Owner

    classDef visitor fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef staff fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef owner fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:3px;
    classDef auth fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;

    class Visitor,Public,PublicChat visitor;
    class Staff,StaffOverview,POS,Transactions staff;
    class Owner,Smartboard,Reports,Branch,Settings owner;
    class Middleware,Session,RoleCheck auth;
    class AI ai;

Role	Surface	Access
Visitor	Homepage + public chat	restaurant information and hospitality chat
Staff	overview subset, cashier, transactions	limited daily operations
Owner	full Smartboard, reports, AI, branch, settings	full decision and administration access

Route Map

flowchart TB
    Root["semayot app"] --> Admin["/admin"]
    Root --> Public["/"]
    Root --> API["/api"]

    Admin --> Login["/admin/login"]
    Admin --> Overview["/admin/overview"]
    Admin --> Menu["/admin/menu"]
    Admin --> POS["/admin/pos"]
    Admin --> Tx["/admin/transactions"]
    Admin --> Reports["/admin/reports"]
    Admin --> Customers["/admin/customers"]
    Admin --> Trends["/admin/customers/trends"]
    Admin --> AIPage["/admin/ai"]
    Admin --> SEO["/admin/seo"]
    Admin --> Settings["/admin/settings"]

    API --> PublicChat["/api/chat"]
    API --> AdminAPI["/api/admin"]
    AdminAPI --> OverviewSummary["/api/admin/overview/summary"]
    AdminAPI --> AIChat["/api/admin/ai/chat"]
    AdminAPI --> AISummary["/api/admin/ai/summary"]
    AdminAPI --> AIMemory["/api/admin/ai/memory"]
    AdminAPI --> Branches["/api/admin/branches"]
    AdminAPI --> MenuAPI["/api/admin/menu"]
    AdminAPI --> POSAPI["/api/admin/pos"]
    AdminAPI --> TxAPI["/api/admin/transactions"]
    AdminAPI --> ReportsDaily["/api/admin/reports/daily"]
    AdminAPI --> ReportsMonthly["/api/admin/reports/monthly"]
    AdminAPI --> ReportsExport["/api/admin/reports/export"]

    classDef root fill:#111827,stroke:#FF4F79,color:#FFFFFF,stroke-width:3px;
    classDef public fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef admin fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:2px;
    classDef api fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef reports fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;

    class Root root;
    class Public,PublicChat public;
    class Admin,Login,Overview,Menu,POS,Tx,Reports,Customers,Trends,SEO,Settings admin;
    class API,AdminAPI,OverviewSummary,Branches,MenuAPI,POSAPI,TxAPI api;
    class AIPage,AIChat,AISummary,AIMemory ai;
    class ReportsDaily,ReportsMonthly,ReportsExport reports;

Admin Workspace

Smartboard is the decision center. Workspace pages are the execution surfaces.

Route	Function	Relation to Smartboard
`/admin/menu`	manage menu	action surface for menu performance and stock signals
`/admin/pos`	cashier and transaction workflow	direct staff execution surface
`/admin/transactions`	transaction history	source for latest activity
`/admin/reports`	business reports	KPI drill-down
`/admin/customers`	customers	basis for retention and trends
`/admin/customers/trends`	customer trends	loyalty and traffic signals
`/admin/ai`	full AI workspace	continuation from overview chat
`/admin/seo`	public site optimization	brand discovery
`/admin/settings`	outlet configuration	operational control

Public Site

The public site remains the brand front door for Semayot, but it is not the product's center of gravity.

Public Surface	Function
Homepage	introduces the restaurant brand
Hero media	builds taste and visual identity
Restaurant info	communicates outlet information
Visitor chat widget	customer conversation
`llms.txt`	information surface for AI crawlers/agents

Owner AI Summary Flow

sequenceDiagram
    participant Owner as Owner
    participant AIView as /admin/ai
    participant Route as /api/admin/ai/summary
    participant Agg as lib/admin/ai/aggregate.ts
    participant Prompt as lib/admin/ai/prompts.ts
    participant Gemini as Gemini
    participant DB as Supabase ai_summaries

    Owner->>AIView: select period today / 7d / 30d
    AIView->>Route: request summary
    Route->>DB: check period cache
    alt cache available
        DB-->>Route: cached summary
        Route-->>AIView: render cached result
    else no cache
        Route->>Agg: read transactions + expenses
        Agg-->>Route: aggregate context
        Route->>Prompt: format owner prompt
        Prompt-->>Route: bounded prompt
        Route->>Gemini: generate summary
        Gemini-->>Route: highlights, concerns, actions
        Route->>DB: store summary
        Route-->>AIView: render new summary
    end

Owner Chat Flow

sequenceDiagram
    participant Owner as Owner
    participant Brief as SEMA Daily Brief
    participant Chat as OverviewChatPanel
    participant Route as /api/admin/ai/chat
    participant Memory as AI Memory
    participant Context as Business Context
    participant Gemini as Gemini

    Brief->>Owner: show quick prompt
    Owner->>Chat: click quick prompt / type question
    Chat->>Route: stream chat request
    Route->>Memory: read unread memory
    Route->>Context: include business summary
    Route->>Gemini: generate bounded answer
    Gemini-->>Route: streaming response
    Route-->>Chat: stream answer
    Chat-->>Owner: operational recommendation dialogue

Stock Risk Radar

flowchart LR
    Inventory["Inventory Items"] --> Classifier["Risk Classifier"]
    Classifier --> Critical["critical<br/>stock requires action"]
    Classifier --> Warning["warning<br/>stock needs monitoring"]
    Classifier --> Safe["safe<br/>stock is sufficient"]
    Classifier --> Unknown["unknown<br/>not enough data"]

    Critical --> Priority["Today's Priorities"]
    Warning --> Priority
    Safe --> Smartboard["Smartboard Display"]
    Unknown --> Empty["Honest Unknown State"]
    Priority --> Smartboard
    Empty --> Smartboard

    classDef source fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef critical fill:#FEE2E2,stroke:#DC2626,color:#7F1D1D,stroke-width:3px;
    classDef warning fill:#FEF3C7,stroke:#D97706,color:#78350F,stroke-width:2px;
    classDef safe fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;
    classDef unknown fill:#F8FAFC,stroke:#64748B,color:#334155,stroke-width:2px;
    classDef board fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:3px;

    class Inventory,Classifier source;
    class Critical critical;
    class Warning warning;
    class Safe safe;
    class Unknown,Empty unknown;
    class Priority,Smartboard board;

Readiness State Model

stateDiagram-v2
    [*] --> Loading
    Loading --> Live: valid data
    Loading --> Demo: demo mode
    Loading --> Empty: insufficient data
    Loading --> Error: request failed

    Live --> SmartboardOperational
    Demo --> SmartboardDemo
    Empty --> SmartEmptyState
    Error --> SafeFailure

    SmartboardOperational --> ActionLoop
    SmartboardDemo --> EducationLoop
    SmartEmptyState --> SetupGuidance
    SafeFailure --> RetryOrEscalate

    ActionLoop --> [*]
    EducationLoop --> [*]
    SetupGuidance --> [*]
    RetryOrEscalate --> [*]

Architecture Folder Map

semayot/
├── app/
│   ├── admin/
│   │   ├── (authenticated)/        # active admin pages
│   │   ├── login/                  # admin login
│   │   └── layout.tsx              # admin wrapper
│   ├── api/
│   │   ├── admin/
│   │   │   ├── ai/                 # chat, summary, memory
│   │   │   ├── overview/summary/   # Smartboard loader
│   │   │   ├── reports/            # daily, monthly, export
│   │   │   ├── menu/
│   │   │   ├── pos/
│   │   │   ├── inventory/
│   │   │   ├── staff/
│   │   │   ├── customers/
│   │   │   └── branches/           # add new outlet
│   │   └── chat/                   # public customer chat
│   ├── globals.css
│   ├── layout.tsx
│   └── page.tsx
├── components/
│   ├── admin/
│   │   ├── overview/               # heart of Sentra Smartboard
│   │   └── pages/                  # admin work pages
│   ├── semayot/                    # public restaurant surface
│   └── ui/                         # utility components
├── lib/
│   ├── admin/
│   │   ├── ai/                     # prompts + AI aggregation
│   │   ├── overview/               # contracts, adapter, intelligence
│   │   ├── schemas/                # admin validation
│   │   ├── supabase/               # client + generated types
│   │   └── rls/                    # SQL/RLS artifacts
│   ├── http/                       # response/error helpers
│   └── semayot/                    # public copy and knowledge
├── public/
│   ├── semayot/images/             # public assets in use
│   ├── Chat WhatsApp.lottie
│   └── llms.txt
└── tests/
    ├── admin/overview/
    ├── admin/routes/
    ├── admin/reports/
    └── http/

Dependency Map

flowchart LR
    Next["Next.js 16<br/>App Router"] --> App["Semayot App"]
    React["React 19"] --> App
    TS["TypeScript strict"] --> App
    Tailwind["Tailwind CSS v4"] --> UI["UI System"]
    Motion["Framer Motion"] --> UI
    Supabase["Supabase SSR + browser client"] --> Data["Auth + Data Layer"]
    AISDK["AI SDK"] --> AI["Gemini Integration"]
    Gemini["Google Gemini"] --> AI
    Vitest["Vitest"] --> Tests["Test Suite"]

    App --> UI
    App --> Data
    App --> AI
    App --> Tests

    classDef framework fill:#111827,stroke:#FF4F79,color:#FFFFFF,stroke-width:3px;
    classDef ui fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:2px;
    classDef data fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef test fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;

    class Next,React,TS,App framework;
    class Tailwind,Motion,UI ui;
    class Supabase,Data data;
    class AISDK,Gemini,AI ai;
    class Vitest,Tests test;

Current Stack

Layer	Technology
Framework	Next.js 16 App Router
Runtime UI	React 19
Language	TypeScript strict
Styling	Tailwind CSS v4
Motion	Framer Motion
Auth/Data	Supabase SSR + browser client
AI	AI SDK + Google Gemini
Test	Vitest
Package manager	pnpm only

Local Runbook

pnpm dev
pnpm build
pnpm start
pnpm lint
pnpm test

Execution notes:

Use pnpm, not npm/yarn/bun.
Admin AI and public chat require GEMINI_API_KEY or GOOGLE_GENERATIVE_AI_API_KEY.
Supabase is used for auth, transactions, inventory, reports, and memory.
Smartboard must remain safe when data is empty: show empty or unknown, not fake numbers.

Environment Variables

Variable	Required By	Notes
`GEMINI_API_KEY`	AI SDK / Gemini	used for public chat and owner AI
`GOOGLE_GENERATIVE_AI_API_KEY`	alternative Gemini key	fallback according to implementation
`NEXT_PUBLIC_SUPABASE_URL`	Supabase client	Supabase project URL
`NEXT_PUBLIC_SUPABASE_ANON_KEY`	browser client	anon key governed by RLS
`SUPABASE_SERVICE_ROLE_KEY`	server-only operations	must never be exposed to client
`NEXT_PUBLIC_APP_URL`	app runtime	base URL for routes/callbacks when needed

Actual variable names must follow the current .env.example or deployment environment. This table is an architectural checklist, not a replacement for secret configuration.

Testing and Governance

flowchart TD
    Change["Code Change"] --> Typecheck["TypeScript Check"]
    Change --> Unit["Vitest Unit Tests"]
    Change --> Route["Route Contract Tests"]
    Change --> UI["Smartboard UI Review"]
    Change --> Guardrail["Data Truth Guardrail"]

    Typecheck --> Pass["Can Merge"]
    Unit --> Pass
    Route --> Pass
    UI --> Pass
    Guardrail --> Pass

    Typecheck -. fail .-> Block["Block Merge"]
    Unit -. fail .-> Block
    Route -. fail .-> Block
    UI -. fail .-> Block
    Guardrail -. fail .-> Block

    classDef change fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef check fill:#E0F2FE,stroke:#0284C7,color:#0C4A6E,stroke-width:2px;
    classDef pass fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:3px;
    classDef block fill:#FEE2E2,stroke:#DC2626,color:#7F1D1D,stroke-width:3px;

    class Change change;
    class Typecheck,Unit,Route,UI,Guardrail check;
    class Pass pass;
    class Block block;

Required Validation

Area	Minimum Check
Type safety	TypeScript strict must not break
Overview contract	adapter and contract must not mismatch
Intelligence	scoring and priorities must remain deterministic
API routes	auth/session/role checks must not rely on middleware alone
Empty state	missing data must be displayed honestly
AI output	must be role-aware and context-bound
UI	Smartboard remains the primary surface, not a generic admin dashboard

Security and Data Boundary

flowchart LR
    Client["Client UI"] --> Middleware["Middleware Gate"]
    Middleware --> AdminRoute["Admin Route Handler"]
    AdminRoute --> Session["Session Re-check"]
    Session --> Role["Role Authorization"]
    Role --> Supabase["Supabase + RLS"]
    Supabase --> Summary["Server Summary"]
    Summary --> Client

    Client -. no direct privileged write .-> Supabase
    Public["Public Visitor"] --> PublicRoute["/api/chat"]
    PublicRoute --> PublicPrompt["Public Prompt Boundary"]
    PublicPrompt --> Gemini["Gemini"]

    Owner["Owner"] --> OwnerAIRoute["/api/admin/ai/**"]
    OwnerAIRoute --> OwnerRole["Owner-only Check"]
    OwnerRole --> Gemini

    classDef client fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:2px;
    classDef auth fill:#F5F3FF,stroke:#8B5CF6,color:#2D1B69,stroke-width:2px;
    classDef data fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:2px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;
    classDef public fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;

    class Client,Owner client;
    class Middleware,AdminRoute,Session,Role,OwnerAIRoute,OwnerRole auth;
    class Supabase,Summary data;
    class Public,PublicRoute,PublicPrompt public;
    class Gemini ai;

Principles:

Middleware protects /admin/* and /api/admin/*.
Route handlers still re-check session.
The owner role is required for sensitive features such as owner AI, reports, and branch switching.
Smartboard reads outlet summaries from the server summary route instead of making direct client-side claims.
Generated Supabase types are the primary schema typing source.
Server secrets must never enter the client bundle.

Most Important Files

File	Why It Matters
`components/admin/overview/overview-page-client.tsx`	main Sentra Smartboard shell
`components/admin/overview/overview-chat-panel.tsx`	owner/staff consultation panel
`lib/admin/overview/intelligence.ts`	scoring, priorities, stock risk
`lib/admin/overview/outlet-contracts.ts`	Smartboard data contract
`lib/admin/overview/outlet-adapter.ts`	maps summary into Smartboard shape
`app/api/admin/overview/summary/route.ts`	main dashboard data loader
`app/api/admin/ai/chat/route.ts`	owner AI chat
`app/api/admin/ai/summary/route.ts`	owner AI summary
`app/api/admin/ai/memory/route.ts`	AI memory inbox
`components/admin/pages/AIView.tsx`	full AI workspace
`components/admin/Sidebar.tsx`	admin navigation
`components/admin/Topbar.tsx`	top-level admin context

Product Truth Principles

Semayot Smartboard is intentionally governed by these product rules:

Do not invent numbers. If data is missing, show an empty state.
Separate real data, deterministic inference, and generative output.
Deterministic intelligence is the backbone of the decision layer. LLMs explain, summarize, and support dialogue.
Owner AI is a business copilot, not a replacement for owner judgment.
Public and admin surfaces may share a codebase, but their context and guardrails must stay different.
Smartboard must be actionable. Every important signal should lead to a next action.
Unknown is safer than a fake number.
The command center must be concise, honest, and fast to read.

Current Product Center of Gravity

flowchart TD
    Homepage["Homepage"] --> Smartboard["Sentra Smartboard"]
    PublicChat["Public Chat"] --> Smartboard
    AdminPages["Admin Workspace"] --> Smartboard
    Reports["Reports"] --> Smartboard
    POS["POS"] --> Smartboard
    Inventory["Inventory"] --> Smartboard
    AI["Owner AI"] --> Smartboard
    Memory["Memory Inbox"] --> Smartboard

    Smartboard --> Decision["Owner Decision"]
    Decision --> Action["Operational Action"]
    Action --> NewData["New Outlet Data"]
    NewData --> Smartboard

    classDef feeder fill:#FFF7D6,stroke:#D97706,color:#422006,stroke-width:2px;
    classDef board fill:#FFE8EE,stroke:#FF4F79,color:#4A1020,stroke-width:4px;
    classDef action fill:#ECFDF5,stroke:#10B981,color:#065F46,stroke-width:3px;
    classDef ai fill:#FFF1F2,stroke:#FB7185,color:#881337,stroke-width:2px;

    class Homepage,PublicChat,AdminPages,Reports,POS,Inventory feeder;
    class AI,Memory ai;
    class Smartboard board;
    class Decision,Action,NewData action;

To understand Semayot today, do not start from the homepage.

Start from Sentra Smartboard.

That is where all important decisions converge:

operational numbers
outlet health
stock risk
cashier closing
daily priorities
AI summaries
owner conversation with SEMA
follow-up action into inventory, cashier, reports, and AI workspace

LET'S CONNECT

INSTRUMENTATION