π Our Project Story
Inspiration
We all love buying cool stuff. The problem? Weβre not as good at letting go.
That exercise bike, those old phones, the kitchen gadgets β they pile up until our homes look more like warehouses. And when we finally decide to sell, the grind begins:
- Take 12 pictures of the same toaster.
- Write a catchy description.
- Post to 4 different platforms.
- Answer 15 messages that all say: βIs this still available?β
Itβs boring, repetitive, and honestlyβ¦ most of us just give up. Stuff stays. Value is lost.
As computer science people, we thought:
What if selling stuff was as easy as throwing it away β but instead of losing money, you make money?
That was the spark for deCluttered.ai .
π What We Built
An end-to-end automated resale assistant that takes you from clutter β cash in a few taps.
- Camera Scan β Record your room or upload a photo.
- AI Object Detection β Items are identified and sorted into resellable vs. non-resellable.
- Smart Pricing β AI sets fair, optimized resale prices.
- Marketplace Creation β Listings are auto-generated (title, description, photos, tags).
- AgentMail Responses β Handles repetitive buyer questions automatically.
- Offer Tracking β Surfaces the best offers so you donβt waste time bargaining.
You just say yes/no to which items to sell β we handle the rest.
β Tech Stack
- Frontend/UI β Next.js, TailwindCSS, Framer Motion
- Backend β Node.js, Express
- AI/ML Models β
- Object detection: YOLO / Detectron2
- Pricing: custom resale pricing model
- Object detection: YOLO / Detectron2
- Automation Agents β
- Fetch.ai β autonomous workflows & marketplace automation
- AgentMail β auto-replies + negotiation assistant
- Fetch.ai β autonomous workflows & marketplace automation
- Database β PostgreSQL / MongoDB
- Hosting β Vercel + AWS
Challenges We Faced
- Teaching AI to spot whatβs worth selling in a messy room.
- Making the UI feel simple and decluttered, not βAI-overwhelming.β
- Building marketplace listing creation automation instead of simple syncing.
- Automating buyer messages without losing the human feel.
- Keeping all of this running hands-free but trustworthy.
π What We Learned
- Automation works best when it feels invisible and intuitive.
- A calm, clean design builds more trust than flashy AI gimmicks.
- Decluttering isnβt just technical β itβs psychological. If it feels easy, people will actually do it.
- Agents like Fetch.ai + AgentMail are powerful building blocks for real-world automation.
π Next Steps
- More marketplace integrations (Craigslist, Facebook Marketplace, etc.).
- Smarter pricing with real-time demand tracking.
- Expanded auto-reply templates for better buyer interaction.
- A seller insights dashboard (earnings, time saved, most valuable items).
- Gamification: streaks, progress milestones, βroom transformation before/after.β
β¨ In one line:
deCluttered.ai makes selling your old stuff as effortless as throwing it away β except this time, your trash pays you back.

Log in or sign up for Devpost to join the conversation.