AI Task Viewer

WhatsApp connected with AI, which can process voice messages as well to create/update/delete tasks. All the tasks are stored in the database of Supabase which is connected to the project.

Comment

Inspiration

Most task management tools live in apps people rarely open, while real work conversations happen on WhatsApp. At the same time, many AI-powered task systems try to automate everything and become unreliable. I wanted a better balance—AI that understands intent, automation that executes safely, and a clean interface that simply shows accurate task information.

What it does

AI Task Viewer is a read-only, AI-powered task dashboard connected to WhatsApp via n8n. I can send natural language messages to ask about tasks or request updates.

The AI:

  • Returns detailed task information when I ask for details
  • Responds with exactly updating when I intend to change a task, triggering automation in n8n

Tasks are displayed in a responsive dashboard with clear status, priority, and due dates, updated in real time through webhooks.

How I built it

  • Frontend: React single-page application with a responsive, card-based UI
  • AI: OpenRouter LLM for intent detection and task understanding
  • Automation: n8n workflows connecting WhatsApp messages to the system
  • State Management: In-memory task state updated only through webhook data
  • Design: Tailwind CSS with gradients, accessibility, and smooth micro-interactions

A strict system prompt keeps the AI deterministic and safe—especially for updates.

Challenges I ran into

  • Accurately distinguishing between task detail requests and update intents
  • Preventing the AI from performing unauthorized or partial updates
  • Handling real-time webhook updates and loading states smoothly
  • Ensuring proper validation, security, and error handling

These challenges pushed me to favor simplicity and reliability over complexity.

Accomplishments that I'm proud of

  • Designing a deterministic AI control pattern using a single keyword: updating
  • Building a seamless WhatsApp → AI → n8n → UI flow
  • Creating a safe, read-only AI system that avoids unintended side effects
  • Delivering a polished, production-ready UI as a solo developer

What I learned

  • AI works best as an intent interpreter, not an executor
  • Clear system prompts dramatically improve AI reliability
  • Simple, deterministic systems scale better than over-automated ones
  • Combining AI with automation platforms enables powerful real-world workflows

What's next for AI Task Viewer

  • Task filtering, search, and analytics
  • Multi-language and voice message support
  • Team and role-based views
  • Export options and optional persistence

My goal is to make messaging apps a reliable, first-class interface for task intelligence, without sacrificing control or trust.

Built With

  • n8n
  • nextjs
  • openrouter
  • supabase
  • whatsapp
Share this project:

Updates