Building Astuto Next：加入 AI 重複建議偵測的公開意見平台

🎃 Building Astuto Next – A Public Feedback Board with AI-Powered Duplicate Detection

I've been working on a new project called Astuto Next, inspired by the idea of making product feedback easier to share, discover, and manage.

The basic problem I wanted to explore was this: when users have ideas for improving a product, their suggestions can get scattered across different platforms. Sometimes, multiple people even suggest the same feature without knowing that someone else has already requested it.

So, I wanted to build a simple platform where people can share ideas, discover existing requests, and vote for the features they would actually like to see.

That's how Astuto Next started! 💜

✨ What is Astuto Next?

Astuto Next is a public feedback board where users can explore companies, submit feature requests, and support ideas through community voting.

Instead of feedback getting lost, the platform brings suggestions together in one place.

🚀 Features I've built

🏢 1. Company Feedback Boards

Users can add companies and explore their feedback boards. Each company has its own space where people can share suggestions and view existing requests.

💡 2. Feature Requests

Users can submit their ideas by adding a title and description. This makes it easier for others to understand the problem and the improvement being suggested.

👍 3. Community Voting

Not every suggestion needs to be submitted again. Users can browse existing ideas and vote for the ones they want companies to consider. This also helps popular requests get noticed.

🤖 4. AI Duplicate Finder

This is one of the features I was particularly excited to experiment with!

Before submitting a new request, users can ask AI to check whether a similar suggestion already exists for that company.

I'm using a local Ollama model (Llama 3.2 1B) for this. It compares the new request with existing suggestions and identifies possible matches.

If a similar request is found, users can support the existing idea instead of creating another one.

🎃 5. Halloween-inspired Interface

I also experimented with the visual design, giving the platform a dark Halloween-inspired theme with purple and orange colours.

🛠️ Tech Stack

Here are the technologies I used:

HTML – Page structure
CSS – Styling and responsive design
JavaScript – Frontend interactions and API requests
Node.js – Backend runtime
Express.js – API routes and server
SQLite (better-sqlite3) – Storing companies and feature requests
Ollama + Llama 3.2 1B – Local AI integration

🧠 Challenges and What I Learned

This project was a good learning experience because I got to work on both frontend and backend functionality.

Some things I learned and explored while building it:

Connecting frontend actions with backend API routes.
Creating and managing database tables using SQLite.
Fetching and displaying data dynamically.
Implementing voting functionality.
Connecting a local AI model with an Express backend.
Handling API errors and debugging issues when the frontend and backend didn't communicate as expected.

The AI integration was especially interesting. Getting Ollama to respond successfully was one part, but making it compare feature requests correctly was another challenge.

I also learned that an AI feature needs proper testing rather than simply displaying an AI-generated response.

🔮 What's Next?

Astuto Next is still a work in progress, and I have more ideas for it.

Some improvements I'd like to explore include:

AI-powered feedback analysis to identify common topics.
Notifications when a feature request changes status.
Better ways for companies to interact with community feedback.
Improving duplicate detection and its accuracy.

💭 Final Thoughts

I started this project to explore how an existing open-source idea could be extended with AI and turned into something useful.

It wasn't just about adding AI to a website. I wanted to understand how it could solve an actual problem for users.

There were definitely a few debugging struggles along the way, but getting the different parts connected and working made the process worth it.

Github Repository - https://github.com/nailaAkr/astuto-nexxt

I'd love to hear your thoughts!

What other AI-powered features would you add to a public feedback platform?

Thanks for reading! 💜