FutureFlow AI — Enterprise Future Process Designer

FutureFlow AI is an enterprise AI solution designed for the Modus Enterprise AI Build Challenge. It takes complex operational workflows, extracts structured steps, diagnoses bottlenecks, maps specific AI opportunities, and synthesizes a future-state workflow with clear human | ai | hybrid division of responsibility and quantified ROI benefits.

---

⚡ Single-Command Evaluation (Docker Compose)

The entire stack (PostgreSQL database + Express API + Nginx Web Frontend) runs with one command.

1. Configure Environment

cp .env.example .env

Edit .env and set your GEMINI_API_KEY:

GEMINI_API_KEY="your-gemini-api-key"

2. Start the Stack

docker compose up --build

• Web UI: http://localhost:3000
• Backend API: http://localhost:4000/health
• Schema migrations run automatically on startup via prisma migrate deploy.

---

🛠️ Local Development Setup (Manual)

1. Prerequisites

• Node.js >= 18
• pnpm (npm install -g pnpm)
• PostgreSQL database
• Gemini API Key

2. Environment Configuration

Create .env in apps/api/.env:

PORT=4000
DATABASE_URL="postgresql://username:password@localhost:5432/futureflow?schema=public"
LLM_PROVIDER="gemini"
# Choose any Gemini model (e.g. gemini-2.5-flash, gemini-2.0-flash, gemini-1.5-pro, etc.)
GEMINI_MODEL="gemini-2.5-flash"
GEMINI_API_KEY="your-gemini-api-key-here"

3. Install & Run

# Install dependencies across workspace
pnpm install

# Run database migrations
pnpm --filter @futureflow/api prisma:migrate

# Start Backend (:4000) and Frontend (:3000)
pnpm dev

---

🔄 LLM Provider Flexibility

The pipeline interacts strictly through generateStructured(prompt, schema) defined in apps/api/src/llm/provider.ts and is decoupled from the underlying vendor.

---

📚 Documentation

• Architecture & Sequence Flow
• Data Model & ER Diagram

---

📦 Third-Party Libraries & Licenses

All third-party open-source libraries used in this project are listed below with their respective software licenses:

Layer / Package	Library	License	Purpose
Backend API	express	MIT	HTTP server and REST routing
	@google/genai	Apache-2.0	Official Google GenAI SDK for Gemini LLM
	@prisma/client / prisma	Apache-2.0	Type-safe ORM & database migrations
	zod	MIT	Runtime schema validation & structured JSON parsing
	cors	MIT	Cross-Origin Resource Sharing middleware
	dotenv	BSD-2-Clause	Environment variable management
	tsx	MIT	TypeScript runtime execution for development & scripts
Web Frontend	react / react-dom	MIT	UI component library and virtual DOM
	tailwindcss	MIT	Utility-first CSS styling & design system
	vite	MIT	Next-generation frontend bundler & dev server
	@vitejs/plugin-react	MIT	React Fast Refresh Vite plugin
	postcss / autoprefixer	MIT	CSS transformations & vendor prefixing
Infrastructure	PostgreSQL 16	PostgreSQL License (Permissive)	Relational persistence store
	Nginx	2-Clause BSD	Static frontend web server & reverse proxy
	Node.js 20	MIT / Node License	JavaScript runtime environment

---

🤖 AI Coding Assistant Disclosure

In compliance with challenge transparency guidelines:

• AI Coding Assistants Used: Google Antigravity was utilized during the development of this project for pair programming.
• Human Oversight & Verification: All domain logic (the 5-step re-engineering pipeline, enterprise ROI benefit calculation heuristics, deterministic schema validation safeguards, and database normalization) was reviewed, audited, and tested for enterprise reliability and correctness.
• Model Decoupling: Application code interacts with AI capabilities strictly through standard typed provider interfaces (LLMProvider), ensuring vendor neutrality and modularity.