KindMinds AI

KindMinds is a conversational AI web application designed for supportive, reflective interaction rather than pure question‑answering.

Instead of acting like a search engine, the system focuses on structured dialogue — helping users articulate thoughts, reflect, and explore ideas through guided conversation.

The project explores how interface design + conversation constraints can shape healthier human‑AI interaction.

---

What it does

• chat‑based interaction with memory
• structured conversational flow (not unrestricted prompting)
• state‑aware responses
• message synchronization across sessions
• controlled AI behavior (see restrictions)

This is not intended to replace professional help — the system is designed to be assistive, not authoritative.

---

Tech Stack

Frontend

• Next.js (App Router)
• TypeScript
• Tailwind
• shadcn/ui

Backend

• Node API routes
• Python AI processing layer

Data

• Supabase (auth + database + realtime)
• Postgres

Infrastructure

• Docker support
• environment‑based config

---

Architecture Overview

client (Next.js)
      ↓
api routes
      ↓
conversation manager
      ↓
AI response layer (python)
      ↓
database (supabase)

The application separates:

• UI interaction
• conversation state logic
• AI generation
• persistence

This allows experimenting with different response models without rewriting the interface.

---

Local Setup

1. Clone

git clone https://github.com/aakashbisht0704/kind-minds-ai.git
cd kind-minds-ai

2. Environment

Create .env.local using the sample:

cp env.sample .env.local

Fill required Supabase and AI keys.

---

3. Install dependencies

npm install

---

4. Run

npm run dev

Open http://localhost:3000

---

Docker (optional)

docker build -t kindminds .
docker run -p 3000:3000 kindminds

---

Project Structure

src/                → frontend + routes
backend/            → AI handling
supabase/           → database migrations
deploy/             → deployment config
public/             → static assets

---

Design Goals

This project experiments with:

• safer conversational AI patterns
• bounded responses instead of open‑ended generation
• persistent conversational context
• UI‑driven behavior control

The focus is understanding interaction design around AI — not just model output quality.

---

Status

Active experiment — behavior, prompts, and constraints change frequently.

---

Notes

See:

• AI_RESTRICTIONS.md
• CHAT_FEATURES.md
• KINDMINDS_CONTEXT.md