Type a topic and get a finished, editable slide deck that you can present, share or export. An LLM-backed SaaS I built from idea to live product.
🌐 Live app: https://presentpitch.vercel.app
PresentPitch turns a short prompt into a complete presentation. It plans the outline, writes the slides and lays them out as editable blocks, then lets you refine, present, share and export the result.
- AI deck generation: prompt in, structured slides out (OpenAI GPT-4o by default, with Groq and Gemini support in the codebase).
- "Thinking" mode: a multi-step agent pipeline (planner → research → generator → critic) runs as a background job for higher-quality decks.
- AI editing helpers: shorten, expand, rewrite or change the tone of text, and generate speaker notes.
- Block-based editor: headings, lists, images, code, quotes and tables, with drag-and-drop ordering and themes.
- Real-time collaboration: live presence and editing over WebSockets (Socket.io).
- Presentation mode: full screen with keyboard navigation, plus share links for viewers.
- Export to PPTX, PDF and HTML.
- Accounts and billing: email/password and Google OAuth sign-in, with Stripe subscription plans.
- Engineering: NestJS + Prisma API, Redis-backed job queues, Jest tests and GitHub Actions CI, Docker files for local and production use.
| Layer | Tech |
|---|---|
| Frontend | Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS, TanStack Query, Zustand, Radix UI, dnd-kit, Socket.io client |
| Backend | NestJS 11, Prisma, PostgreSQL, Redis, Socket.io, Passport (JWT + Google OAuth) |
| AI | OpenAI API (GPT-4o / GPT-4o-mini), Groq, Google Gemini |
| Payments and infra | Stripe, AWS S3, Docker, Vercel (frontend), Render (backend) |
You'll need: Node.js 24.12+ (see .nvmrc), Docker (for Postgres and Redis) and an OpenAI or Groq API key.
git clone https://github.com/Faizanmal/presentation_generator.git
cd presentation_generator
# 1. Environment files
cp .env.example .env
cp backend-nest/.env.example backend-nest/.env
cp frontend/.env.example frontend/.env.local
# then set DATABASE_URL, REDIS_*, and OPENAI_API_KEY (or GROQ_API_KEY) in backend-nest/.env
# 2. Database and cache
docker compose up -d db redis
# 3. Backend (http://localhost:3001)
cd backend-nest
npm install
npx prisma generate
npx prisma migrate dev
npm run start:dev
# 4. Frontend (http://localhost:3000), in a second terminal
cd frontend
npm install
npm run devThe frontend reads NEXT_PUBLIC_API_URL (default http://localhost:3001/api).
Tests: run npm test in backend-nest/ or frontend/.
frontend/ # Next.js app (dashboard, editor, present, view, settings)
backend-nest/ # NestJS API (ai, slides, blocks, export, collaboration, payments, auth, ...)
shared/ # shared presentation DSL and layout engine
docs/ # deployment and feature guides
I build LLM features into real products, from an AI assistant or RAG MVP to a full SaaS like this one, in fixed scopes with clear milestones, tests and CI. I also stay on after launch for maintenance.
Faizan Malik, Freelance AI & Automation Engineer (Python, Next.js, LLM/RAG), based in Delhi (IST, UTC+5:30) → Contra · LinkedIn · GitHub
MIT