Giathi-Daniel/kenya-finance-bill-2026-ai

The project contains code and resources for a modern AI-powered Finance Bill chatbot built with Next.js, Groq, and the AI SDK. It provides a streaming chat interface designed to help users interact with Kenya Finance Bill content through a fast, responsive, and context-aware assistant. # 🇰🇪 Kenya Finance Bill 2026 AI Assistant A free, fast, and open-source AI chatbot designed to explain the **Kenya Finance Bill, 2026** in plain English. Built with Next.js, Vercel, and Groq (Llama 3.2), this tool helps citizens, business owners, and students understand complex tax amendments without needing a law degree. > **What it does:** You ask a question (e.g., *"How does the bill affect gambling winnings?"*), and the AI scans the official 46-page bill to give you a simple answer with specific Section citations. ## Features - **Plain English Explanations:** Translates legalese into simple terms. - **Section Citations:** Every answer references the specific Section (e.g., *Section 12H*, *Section 23(y)*) for verification. - **Zero Cost:** Uses a free-tier API (Groq) and free hosting (Vercel). - **Fast & Serverless:** Built on Next.js App Router with streaming responses. - **Mobile Friendly:** Responsive design works perfectly on phones. ## Quick Start ### Prerequisites - Node.js (v18+) - A free Groq Cloud API key ### 1. Clone the Repository ```bash git clone github.com cd kenya-finance-bill-2026-ai ``` ### 2. Install Dependencies ```bash npm install ``` ### 3. Set Up Environment Variables Create a file named `.env.local` in the root directory: ```env GROQ_API_KEY=gsk_************************** ``` *Get your free key at console.groq.com* ### 4. Run Locally ```bash npm run dev ``` Open `localhost` to see the bot in action. ## How It Works This app uses a **Retrieval-Augmented Generation (RAG)** approach, but simplified for speed: 1. **Context Injection:** The full text of the Finance Bill, 2026 is embedded directly into the client code (`page.tsx`). 2. **Prompt Engineering:** When you ask a question, the system constructs a prompt that includes your question + the full bill text. 3. **LLM Processing:** The prompt is sent to **Groq** (running Llama 3.2), which analyzes the text and gene …