AcademAI — AI Academic Assistant

Live: https://academai-assistant.vercel.app

A study assistant that turns any text or topic into a summary, a level-matched explanation or a quiz. Built with Next.js 16 (App Router), TypeScript and Tailwind CSS, using Groq's OpenAI-compatible API.

What it does

Mode	Options	Output
Summarize	—	Bullet-point summary with key terms in bold and three takeaways
Explain	ELI5 · High School · University	Sectioned explanation with an analogy and a recap
Quiz	5 or 10 questions · easy / medium / hard	Numbered questions followed by answers with short explanations

Plus one-click topic presets, Markdown-rendered output with copy/clear, and a local history of the last 10 requests.

How it's built

Browser (React client)  ──POST /api/ai──▶  Route handler (server only)  ──▶  Groq API
                                             ├─ rate limit (6 req/min per IP)
                                             ├─ Zod validation (1–8,000 chars, valid options)
                                             ├─ safety filter on the input
                                             └─ request ID logged at start / end / error

• The API key never reaches the browser. All model calls happen in src/app/api/ai/route.ts; the client only talks to its own API route.
• Input is validated before it costs anything. Zod checks length and allowed options and returns 400 with a readable message; disallowed requests get 403 before reaching the model.
• Failures are traceable but not leaky. Each request gets a UUID that is logged server-side and returned on errors, while provider error details stay in the logs.
• The model is configurable. GROQ_MODEL overrides the default (openai/gpt-oss-120b), so a provider model retirement is a config change rather than a code change.

Note: the in-memory rate limiter is per server instance, which suits a demo. A multi-instance deployment would move it to a shared store such as Redis.

Tech stack

Next.js 16 · React 19 · TypeScript · Tailwind CSS · Groq API (via the openai SDK) · Zod · react-markdown · Vercel (CI/CD from GitHub)

Run locally

git clone https://github.com/agent-mino/Academai.git
cd Academai
npm install
cp .env.example .env.local   # then add your GROQ_API_KEY (free at https://console.groq.com/keys)
npm run dev                  # http://localhost:3000

Variable	Required	Purpose
GROQ_API_KEY	yes	Groq API key, used server-side only
GROQ_MODEL	no	Override the default model

Scripts

npm run dev     # development server
npm run lint    # ESLint
npm run build   # production build
npm start       # serve the production build

Deployed on Vercel: every push to main builds and deploys automatically.