SmallBox

SmallBox is a self-serve web application that democratizes access to enterprise-grade IBM tools—without requiring technical expertise. Small business owners can now access AI-powered content generation, intelligent financial management, and data-driven marketing insights through an intuitive dashboard.

Real-time AI content generation answers "What should I post?" with one click. Intelligent expense categorization answers "Where is my money going?" with automatic transaction analysis. Sentiment analytics answers "What do customers really think?" with one-tap review analysis. All designed to be simple, private, and always on your side.

The Problem

Small business owners lack access to enterprise tools—their competitors with dedicated IT teams use AI-powered automation, advanced analytics, and intelligent workflow tools. They face three critical challenges:

Time Poverty: Manually writing social posts, categorizing expenses, and analyzing customer feedback consumes hours weekly
Resource Constraints: Can't afford specialized software or hire data analysts
Knowledge Gap: IBM's powerful tools (Watsonx.ai, Watson NLU) are designed for enterprise teams, not small business owners

Existing solutions are fragmented: generic marketing apps don't personalize content; spreadsheets don't automatically categorize transactions; review platforms don't extract actionable insights. SmallBox exists because business intelligence should live in one place, accessible to any business owner.

How It Works

SmallBox combines three core pipelines:

AI Content Generation (Watsonx.ai) — Turn business context into marketing copy
Intelligent Finance (Watsonx.ai + SQLite) — Auto-categorize transactions, track budgets, generate insights
Customer Intelligence (Watson NLU) — Analyze sentiment, extract themes, recommend improvements

All data stays private; credentials are stored securely on the server; the frontend handles UI and display only.

High-Level Flow

flowchart TB
    subgraph input [User Input]
        MARKET["Marketing Form<br/>(Topic, Tone, Platform)"]
        TRANS["Finance Entry<br/>(Expense, Date, Amount)"]
        REVIEW["Review Paste<br/>(Customer Feedback)"]
    end

    subgraph generate [Generation Pipeline]
        WX_CONTENT["Watsonx.ai<br/>Content Generation"]
        WX_CAT["Watsonx.ai<br/>Auto-Categorize"]
        NLU["Watson NLU<br/>Sentiment Analysis"]
    end

    subgraph process [Processing]
        VAR["3 Content<br/>Variations"]
        CAT["Category + Budget<br/>Tracking"]
        INSIGHT["Sentiment Score +<br/>Top Themes"]
    end

    subgraph output [Display & Action]
        CONTENT["Marketing Dashboard<br/>Copy Ready to Use"]
        FINANCE["Finance Dashboard<br/>Budget & Trends"]
        ANALYZE["Review Analytics<br/>Actionable Feedback"]
    end

    subgraph data [Data Layer]
        DB["SQLite<br/>Transactions, Budgets, History"]
    end

    MARKET --> WX_CONTENT --> VAR --> CONTENT
    TRANS --> WX_CAT --> CAT --> FINANCE
    REVIEW --> NLU --> INSIGHT --> ANALYZE
    
    CAT --> DB
    VAR -.->|History| DB
    INSIGHT -.->|History| DB

Marketing Content Pipeline

The Marketing Dashboard combines AI content generation with email campaign management:

Content Generation: User inputs topic (product/service), tone, and platform (Instagram, Facebook, Email, Ad copy)
Watsonx.ai Processing: Sends context to Meta Llama 3.3 70B Instruct model; generates 3 unique variations
Draft Preview: User sees all variations, edits inline, copies to clipboard or schedules send
Email Integration: If email selected, user can parse recipient list (CSV or manual), generate subject + body, and send via Brevo SMTP
Campaign Tracking: Each campaign logged with source (AI or manual), timestamp, and recipient count

Finance Intelligence Pipeline

The Finance Dashboard provides transaction management with AI-powered categorization:

Manual Entry or Upload: User adds expenses/income manually or uploads receipt/statement
Auto-Categorization: Watsonx.ai analyzes description and amount; suggests category (Food, Rent, Utilities, Supplies, etc.)
Budget Tracking: Monthly targets set per category; dashboard shows progress, alerts near/over limit
Insights & Reports: End-of-month summary with spend trends, biggest expense categories, and savings recommendations
Persistent History: All transactions stored in SQLite; searchable, filterable, exportable

Customer Intelligence Pipeline

The Review Analyzer (Marketing tab) extracts value from customer feedback:

Paste Reviews: User inputs customer reviews from Google, Yelp, or manual sources
Sentiment Extraction: Watson NLU processes reviews; returns overall sentiment score (0-100), top positive themes, top complaints
Keyword Cloud: Visualizes most-mentioned terms (positive and negative)
Actionable Feedback: AI suggests 1–2 improvements based on negative themes and complaint patterns
Trend Analysis: If multiple analysis batches submitted, tracks sentiment over time

System Architecture

flowchart TB
    subgraph client [Browser Client]
        REACT["React 19.2.4<br/>+ TypeScript"]
        DASH["Dashboard Pages<br/>Marketing, Finance,<br/>Website, Settings"]
        UI["UI Components<br/>Sidebar, Cards, Forms"]
        REACT --> DASH
        REACT --> UI
    end

    subgraph server [Backend / API Routes]
        NEXT["Next.js 16.2.6<br/>App Router"]
        subgraph api [API Layer]
            CONTENT["POST /api/ibm/watsonx/<br/>generate-content"]
            EMAIL["POST /api/ibm/watsonx/<br/>generate-email"]
            ANALYZE["POST /api/ibm/nlu/<br/>analyze"]
            CAMPAIGN["POST /api/campaigns/<br/>send"]
        end
        NEXT --> CONTENT
        NEXT --> EMAIL
        NEXT --> ANALYZE
        NEXT --> CAMPAIGN
    end

    subgraph lib [Business Logic / Utilities]
        WXUTIL["src/lib/watsonx.ts<br/>- generateContentWithWatsonx<br/>- generateEmailWithWatsonx<br/>- categorizeTransaction<br/>- callWatsonx"]
        NLUUTIL["src/lib/ibm-nlu.ts<br/>- analyzeReviews<br/>- analyzeSentiment"]
        EMAILUTIL["src/lib/email-campaign.ts<br/>- parseEmailRecipients<br/>- createEmailCampaignDraft<br/>- renderEmailCampaignHtml"]
    end

    subgraph external [External APIs]
        WX["IBM Watsonx.ai<br/>Meta Llama 3.3 70B"]
        NLU["IBM Watson NLU<br/>Sentiment Analysis"]
        SMTP["Brevo SMTP<br/>Email Relay"]
    end

    subgraph data [Data Layer]
        DB["SQLite<br/>Transactions, Budgets,<br/>Email History"]
    end

    client <-->|HTTP/REST| server
    CONTENT -.->|fetch| WX
    EMAIL -.->|fetch| WX
    ANALYZE -.->|fetch| NLU
    CAMPAIGN -.->|Nodemailer| SMTP
    CONTENT --> WXUTIL
    EMAIL --> WXUTIL
    ANALYZE --> NLUUTIL
    CAMPAIGN --> EMAILUTIL
    EMAILUTIL --> DB
    WXUTIL --> DB

Frontend Layer

React 19.2.4 + TypeScript + Vite: Component-based UI with server-side rendering via Next.js
Tailwind CSS + Framer Motion: Utility-first styling with smooth animations
Dashboard Pages:
- Marketing — Content generation, email campaigns, review analyzer
- Finance — Transaction entry, budget tracking, analytics, reports
- Website — (v2 planned) Website builder with template system
- Settings — User account management
Sidebar Navigation: Persistent left nav with sections (Home, Marketing, Finance, Website, Settings)
WatsonChat Component: Optional chat interface for natural language queries (lazy-loaded)

Backend / API Routes

Next.js App Router: REST endpoints only (no GraphQL, no WebSockets)
Watsonx Content Generation (POST /api/ibm/watsonx/generate-content): Marketing content with 3 variations
Watsonx Email Generation (POST /api/ibm/watsonx/generate-email): Email draft with subject, body, CTA
Watson NLU Analysis (POST /api/ibm/nlu/analyze): Sentiment analysis + keyword extraction
Email Campaign Send (POST /api/campaigns/send): Nodemailer + Brevo SMTP delivery
Finance CRUD (/api/finance/*): Transaction and budget management

Business Logic & Utilities

src/lib/watsonx.ts:
- callWatsonx(prompt, maxTokens) — Core API caller with fallback templates
- generateContentWithWatsonx(input) — Marketing content (Instagram, Facebook, Email, Ad)
- generateEmailWithWatsonx(input) — Email draft generation
- categorizeTransaction(description, amount) — Auto-categorize expense/income
- categorizeTransactionsBatch(items) — Batch categorization
src/lib/ibm-nlu.ts:
- analyzeReviews(reviews) — Sentiment extraction + keyword analysis
- analyzeSentiment(text) — Per-review sentiment scoring
src/lib/email-campaign.ts:
- parseEmailRecipients(emailList) — Validate + deduplicate email addresses
- createEmailCampaignDraft(input) — Generate draft structure
- renderEmailCampaignHtml(draft) — HTML email template rendering

Data Layer

SQLite Database: Single smallbox.db file, stores:
- User transactions and budgets (Finance)
- Email campaign history (Marketing)
- Review analysis batches (Marketing)
- Website configurations (Website, v2)
- Settings and preferences
Foreign Keys Enabled: PRAGMA foreign_keys enabled on DB connect for referential integrity

Technical Implementation

Content Generation (Watsonx.ai)

Generation runs server-side only; credentials are stored in .env.local and never exposed to the browser. On Marketing → AI Content:

User inputs topic, tone (professional/friendly/bold), platform (Instagram/Facebook/Email/Ad)
Frontend POSTs to /api/ibm/watsonx/generate-content
Backend calls generateContentWithWatsonx() which:
- Builds a detailed prompt with examples and constraints
- Calls Watsonx.ai /text/chat endpoint (Meta Llama 3.3 70B Instruct)
- Parses JSON response (3 variations)
- Returns result with source: "watsonx" or fallback source: "local" if API error
Frontend displays variations in editable text areas; user picks one to use

Watsonx.ai Setup:

Model: Meta Llama 3.3 70B Instruct (multilingual, instruction-tuned)
Region: Canada (Toronto) — ca-tor.ml.cloud.ibm.com
Endpoint: POST /ml/v1/text/chat?version=2024-03-19 (modern endpoint, 2048 token max)
Auth: IAM token exchange to https://iam.cloud.ibm.com/identity/token
Credentials: WATSONX_API_KEY, WATSONX_PROJECT_ID, WATSONX_URL in .env.local

Email Campaign Flow

Email campaigns combine content generation with SMTP delivery:

Draft Generation: Watsonx generates subject + body + preheader (see Content Generation above)
Recipient Parsing: User pastes email list; parseEmailRecipients() validates, deduplicates, returns count
Draft Preview: User reviews email, can edit subject/body inline
Send: POSTs to /api/campaigns/send with recipients + draft
SMTP Delivery: Nodemailer creates transport to Brevo; sends per-recipient copies; tracks accept/reject
Response: Returns {acceptedCount, rejectedRecipients, sentAt, campaignId}

Brevo SMTP Setup:

Provider: Brevo (formerly Sendinblue)
Free Tier: 300 emails/day, requires verified sender
Host: smtp-relay.brevo.com:587 (TLS)
Credentials: SMTP_USER, SMTP_PASS, SMTP_FROM_EMAIL in .env.local
Nodemailer Config: Uses standard SMTP transport with Brevo relay

Finance Auto-Categorization

Transactions are auto-categorized using Watsonx.ai:

User enters expense: amount $49.99, description "Whole Foods"
Frontend calls /api/finance/categorize with description + amount
Backend calls categorizeTransaction(description, amount) which:
- Sends to Watsonx: "Categorize this expense. Description: '{description}', Amount: ${amount}. Categories: Food, Rent, Utilities, Supplies, Other. Return category name only."
- Parses response; validates against allowed categories
- Falls back to "Other" if error or invalid response
Returns category + confidence; frontend displays and allows user to override
User confirms; transaction saved to SQLite with category

Sentiment Analysis (Watson NLU)

Customer reviews are analyzed for sentiment and themes:

User pastes customer reviews into Review Analyzer (Marketing tab)
Frontend calls /api/ibm/nlu/analyze with review text
Backend calls analyzeReviews(reviews) which:
- Sends to Watson NLU API with emotion, sentiment, keywords features
- Extracts overall sentiment score (negative/neutral/positive)
- Identifies top 5-10 keywords with relevance scores
- Groups positive/negative themes
Returns structured response; frontend displays:
- Overall sentiment (% positive/negative/neutral)
- Keyword cloud (size = relevance)
- Top positive themes
- Top complaints + AI suggestions for improvement

Watson NLU Setup:

API Key: IBM_NLU_API_KEY in .env.local
URL: IBM_NLU_URL (e.g., https://api.us-south.natural-language-understanding.watson.cloud.ibm.com)
Version: IBM_NLU_VERSION (e.g., 2021-08-01)
Features: Sentiment, keywords, emotion analysis

Performance Optimizations

Applied three performance passes to reduce dashboard lag:

Global Animation Reduction (50%):
- Reduced animated beam count from 30 to 15 on background canvas
- Added prefers-reduced-motion media query support
- Pause animation when page not visible (visibilitychange listener)
- Lower blur intensity to reduce GPU work
Marketing Page Memoization:
- Added useMemo for email recipient parsing (prevent recalc on keystroke)
- Added useMemo for draft preview calculation
- Prevents unnecessary re-renders during form input
Dashboard Shell Optimization:
- Replaced Framer Motion animations with plain CSS transitions on Sidebar/Header
- WatsonChat component lazy-mounted only when user opens it
- Defers expensive animation component until user interaction

Tech Stack

Layer	Technology	Purpose
Frontend	React 19.2.4, TypeScript, Vite	Component-based UI with type safety
Framework	Next.js 16.2.6 (App Router)	Full-stack React with server components
Styling	Tailwind CSS 4.0	Utility-first CSS, consistent design tokens
Animation	Framer Motion 12.38.0	Smooth transitions and interactions
Content Gen	IBM Watsonx.ai API	Meta Llama 3.3 70B for content + categorization
Sentiment	IBM Watson NLU API	Emotion, sentiment, and keyword extraction
Email Transport	Nodemailer 8.0.7	SMTP client for email delivery
Email Provider	Brevo SMTP	Email relay (300/day free tier)
Finance Data	SQLite	Persistent transaction + budget storage
Charts	Recharts 3.8.1	Finance dashboard visualizations
3D Graphics	Three.js + React Three Fiber	3D rendering (optional effects, v2)
OCR	Tesseract.js 7.0.0	Receipt/document text extraction (v2)
Icons	Lucide React 1.14.0	Clean, consistent icon library

API Design

The backend exposes REST endpoints only. Full OpenAPI docs available at http://localhost:3000/api-docs (when available).

REST Endpoints

Resource	Methods / Endpoints	Description
Marketing Content	`POST /api/ibm/watsonx/generate-content`	Generate 3 content variations (Instagram, Facebook, Email, Ad)
Email Drafts	`POST /api/ibm/watsonx/generate-email`	Generate email draft (subject, body, preheader, CTA)
Email Campaigns	`POST /api/campaigns/send`	Send email campaign to recipient list (SMTP)
Sentiment Analysis	`POST /api/ibm/nlu/analyze`	Analyze customer reviews (sentiment + keywords)
Finance Transactions	`GET /api/finance/transactions`	List user transactions (filterable by date/category)
	`POST /api/finance/transactions`	Create new transaction
	`GET /api/finance/transactions/{id}`	Get transaction details
	`PATCH /api/finance/transactions/{id}`	Update transaction
	`DELETE /api/finance/transactions/{id}`	Delete transaction
Auto-Categorize	`POST /api/finance/categorize`	Auto-categorize single transaction (Watsonx)
Budget Targets	`GET /api/finance/budgets`	List monthly budget targets
	`POST /api/finance/budgets`	Create budget target
	`PATCH /api/finance/budgets/{id}`	Update budget target
Finance Insights	`GET /api/finance/insights`	Generate monthly insights + recommendations
	`GET /api/finance/report`	Export month/year report (PDF or CSV)
Health	`GET /api/health`	Liveness check (status + dependencies)

Project Structure

IBM-UNSA-HACK/
├── docs/                           # Comprehensive documentation
│   ├── spec.md                     # Product specification (features, scope)
│   ├── DESIGN.md                   # Design system (colors, typography, components)
│   ├── website-builder-prompt.md   # Website builder AI prompt engineering
│   └── marketing/
│       ├── WATSONX_INTEGRATION.md  # Watsonx.ai setup + implementation
│       └── SMTP_EMAIL_CAMPAIGNS_PLAN.md # Email + SMTP architecture
│
└── smallbox/                       # Main Next.js application
    ├── src/
    │   ├── app/
    │   │   ├── layout.tsx           # Root layout (Sidebar, Header, WatsonChat)
    │   │   ├── page.tsx             # Home / Dashboard landing
    │   │   ├── globals.css          # Global styles (Tailwind base)
    │   │   │
    │   │   ├── api/                 # Next.js API routes (serverless functions)
    │   │   │   ├── ibm/
    │   │   │   │   ├── watsonx/     # Watsonx.ai endpoints
    │   │   │   │   │   ├── generate-content/route.ts
    │   │   │   │   │   └── generate-email/route.ts
    │   │   │   │   └── nlu/         # Watson NLU endpoints
    │   │   │   │       └── analyze/route.ts
    │   │   │   │
    │   │   │   ├── campaigns/       # Email campaign endpoints
    │   │   │   │   └── send/route.ts
    │   │   │   │
    │   │   │   └── finance/         # Finance management endpoints
    │   │   │       ├── transactions/route.ts
    │   │   │       ├── budgets/route.ts
    │   │   │       ├── categorize/route.ts
    │   │   │       └── insights/route.ts
    │   │   │
    │   │   ├── dashboard/           # Protected dashboard pages
    │   │   │   ├── layout.tsx       # Dashboard shell (Sidebar, Header)
    │   │   │   │
    │   │   │   ├── marketing/
    │   │   │   │   └── page.tsx     # AI Content, Email, Review Analyzer
    │   │   │   │
    │   │   │   ├── finance/
    │   │   │   │   └── page.tsx     # Transactions, Budgets, Analytics, Reports
    │   │   │   │
    │   │   │   ├── website/
    │   │   │   │   └── page.tsx     # Website Builder (v2 planned)
    │   │   │   │
    │   │   │   └── settings/
    │   │   │       └── page.tsx     # Account & Preferences
    │   │   │
    │   │   └── api/
    │   │       └── ibm/
    │   │
    │   ├── components/              # Reusable React components
    │   │   ├── Sidebar.tsx          # Left navigation (Dashboard pages, Settings)
    │   │   ├── DashboardHeader.tsx  # Top bar (Notifications, User Menu)
    │   │   ├── WatsonChat.tsx       # Optional AI chat interface
    │   │   ├── ui/                  # shadcn/ui component library
    │   │   │   ├── avatar.tsx
    │   │   │   ├── badge.tsx
    │   │   │   ├── button.tsx
    │   │   │   ├── card.tsx
    │   │   │   ├── dropdown-menu.tsx
    │   │   │   ├── input.tsx
    │   │   │   ├── progress.tsx
    │   │   │   ├── select.tsx
    │   │   │   ├── separator.tsx
    │   │   │   ├── tabs.tsx
    │   │   │   └── textarea.tsx
    │   │   └── ...other layout components
    │   │
    │   └── lib/                     # Shared utilities & helpers
    │       ├── watsonx.ts           # Watsonx.ai API wrapper (content generation, categorization)
    │       ├── ibm-nlu.ts           # Watson NLU wrapper (sentiment analysis)
    │       ├── email-campaign.ts    # Email utilities (parsing, drafting, rendering)
    │       └── utils.ts             # General utilities (formatting, validation)
    │
    ├── public/                      # Static assets (images, icons, models)
    │
    ├── package.json                 # Dependencies and scripts
    ├── tsconfig.json                # TypeScript configuration
    ├── next.config.ts               # Next.js configuration
    ├── tailwind.config.ts            # Tailwind CSS theme customization
    ├── postcss.config.mjs            # PostCSS plugins (Tailwind)
    ├── eslint.config.mjs            # ESLint rules
    │
    ├── .env.example                 # Environment variables template
    ├── .env.local                   # Actual secrets (git-ignored)
    ├── .gitignore
    └── README.md                    # Development quick-start

Getting Started

Prerequisites

Node.js 18+ and npm
IBM Cloud Account (free tier available at https://cloud.ibm.com)
Optional: Brevo account for email delivery (free tier: 300/day)

Local Development

Clone & Install
```
cd smallbox
npm install
```

Configure Credentials

cp .env.example .env.local

Fill in .env.local with:

# IBM Watsonx.ai
WATSONX_API_KEY=<your_api_key>
WATSONX_PROJECT_ID=<your_project_id>
WATSONX_URL=https://ca-tor.ml.cloud.ibm.com/ml/v1

# IBM Watson NLU
IBM_NLU_API_KEY=<your_nlu_api_key>
IBM_NLU_URL=https://api.us-south.natural-language-understanding.watson.cloud.ibm.com
IBM_NLU_VERSION=2021-08-01

# Email (Brevo SMTP)
SMTP_HOST=smtp-relay.brevo.com
SMTP_PORT=587
SMTP_SECURE=true
SMTP_USER=<your_brevo_smtp_login>
SMTP_PASS=<your_brevo_smtp_password>
SMTP_FROM_EMAIL=<verified_sender_email>

Get IBM Credentials
- Watsonx.ai: https://dataplatform.cloud.ibm.com/wx
  - Create project, get API key from Manage → API Keys, copy Project ID
- Watson NLU: IBM Cloud Console → Create instance → Get API key + URL
- Brevo SMTP: https://app.brevo.com → Settings → SMTP & API → Get SMTP credentials
Run Development Server
```
npm run dev
```
Open http://localhost:3000
Test Features
- Marketing: Dashboard → Marketing → AI Content (generate Instagram posts)
- Email: Dashboard → Marketing → Email Campaigns (generate + send)
- Finance: Dashboard → Finance → Add transaction (auto-categorize)
- Reviews: Dashboard → Marketing → Review Analyzer (paste customer reviews)

Production Build

npm run build
npm run start

Security & Privacy

Credentials Stored Server-Side: API keys (Watsonx, NLU, Brevo) stored only in .env.local; never exposed to browser
Secure HTTPS: For production, run behind HTTPS only (TLS 1.2+)
Database Encryption: SQLite stores transactions; consider full-disk encryption for production
CORS Configuration: Configure CORS for your frontend origin; restrict API routes to authenticated users (v2 roadmap)
Rate Limiting: Implement rate limits on API routes to prevent abuse (v2 roadmap)
Data Retention: Clearly communicate data retention policy to users; provide export/delete functionality (v2 roadmap)

Engineering Practices

Separation of Concerns

Frontend (src/components/, src/app/dashboard/): Handles UI, form input, display, user interactions
API Routes (src/app/api/): Handles HTTP requests, input validation, response formatting
Business Logic (src/lib/): Encapsulates Watsonx, NLU, email, and finance logic; reusable across routes
Credentials: Stored server-side only; never passed to browser

Type Safety

TypeScript 5.x: All frontend code and API routes fully typed
Pydantic equivalent: On API routes, use Zod or similar for request validation
Strong Interfaces: Define request/response contracts; enforce in routes

Graceful Degradation

Watsonx Fallback: If API fails, timeout, or unconfigured → use local templates (reliable, free fallback)
Email Fallback: If SMTP fails → queue for retry or notify user
NLU Fallback: If Watson NLU unavailable → use basic sentiment heuristics or skip analysis
All features work with reduced quality if external APIs unavailable

Testing & Validation

Environment Validation: Routes check isWatsonxConfigured() before attempting API calls
Input Validation: All email, transaction, and review inputs validated before processing
Error Handling: Descriptive error messages returned to frontend; sensitive errors logged server-side only
Build Validation: npm run build validates TypeScript, ESLint, and builds optimized bundle

Roadmap (v2+)

Feature	Description
Unified AI Assistant	Chat interface: "Generate weekly social posts" or "Show me this month's expenses" — Watsonx becomes central orchestrator
Website Builder	Guided wizard → website template → publish to subdomain (IBM Cloud CD pipeline)
Receipt Scanner	Mobile camera → Tesseract.js OCR → auto-extract items/amounts → add to Finance
Inventory Management	Track stock levels, set low-stock alerts, log purchases
Multi-User & Teams	Invite team members, manage permissions, assign roles
Advanced Analytics	Forecasting, trend analysis, predictive budgeting
Integrations	Zapier, IFTTT, native Slack/Discord bots for alerts
Mobile App	React Native or PWA for on-the-go access

Support & Contribution

Issues: File issues with feature requests, bug reports, or documentation improvements
Security: Report security vulnerabilities to security@smallbox.app (not via public issues)
Contributing: Submit PRs with clear descriptions; ensure TypeScript passes and tests pass
Community: Join our Discord for discussion and support (link in repo)

License

MIT License - see LICENSE for details.

Built with love by the SmallBox team, powered by IBM Cloud technologies.

Name		Name	Last commit message	Last commit date
Latest commit History 37 Commits
docs		docs
smallbox		smallbox
.gitignore		.gitignore
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

SmallBox

The Problem

How It Works

High-Level Flow

Marketing Content Pipeline

Finance Intelligence Pipeline

Customer Intelligence Pipeline

System Architecture

Frontend Layer

Backend / API Routes

Business Logic & Utilities

Data Layer

Technical Implementation

Content Generation (Watsonx.ai)

Email Campaign Flow

Finance Auto-Categorization

Sentiment Analysis (Watson NLU)

Performance Optimizations

Tech Stack

API Design

REST Endpoints

Project Structure

Getting Started

Prerequisites

Local Development

Production Build

Security & Privacy

Engineering Practices

Separation of Concerns

Type Safety

Graceful Degradation

Testing & Validation

Roadmap (v2+)

Support & Contribution

License

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages