TemplateAI

(It's kind of reusable prompt system)

• Reusable AI prompt templates
• Eliminating repetitive retyping while maintaining stateful context.
• Will be useful for users who frequently work with structured prompts

Projected Features

Reusable Prompt Templates

• Save prompts once and reuse forever with dynamic placeholders (${variable} syntax).
• Build multiple template sets categorized by project, tone, task, or client.

Stateful Prompt 🧠

• AI sessions remember context using thread-based architecture.
• Prompts can "recall" prior input.

Bring Your Own Key (BYOK)

• Use your own OpenAI API keys.

Folder-Based Thread Organization

• Chat threads can be organized into folders.
• Quickly switch between templates and session states.

Voice-to-Prompt Input (Long Term Vision)

• Integrated speech-to-text via Whisper and Convex speech layer.
• Enables hands-free interaction.

Usage Analytics Dashboard

• Track which prompts are used most, user session length, and token spend.
• Built using Convex analytics functions.

Configurable Settings

• Switch models, memory strategies, search providers, themes, and more.
• Customize user experience per-session or globally.

Projected UI

Projected Tech Stack

Frontend

• React + Vite + TypeScript
• ShadCN/UI components with tailwind-based design system
• Zustand + custom stores for managing prompt state
• Voice & multimodal input via use-voice-recorder.ts + multimodal-input.tsx
• Dynamic theming and layout with theme-provider.tsx

Backend

• Convex as the real-time backend
• Handles threads, messages, attachments, search integrations
• Schema defined with Convex TypeScript DSL
• Authentication via NextAuth with GitHub & Google
• Prompt enhancement pipeline via chat_http/prompt.ts