Inspiration

As a long time renovator, I have often wanted to share my renovation journey with friends and family - where traditional methods were combersome, usually across multiple platforms (facebook, text messages), searching/scrolling through large phone photo albums (losing attention of friends whilst they're politely waiting). Then not to mention the pain and trouble of sourcing reliable contractors/providers - with more trust than with any solid proof of experience or reviews.

There was born the idea Renoview! The idea was conceived in 2008 for my first renovation.. though was facing $40k+ build efforts - which was a tall ask to fund for a hobby project.. so it got shelved.

It wasn't until about October in 2025 - when I challenged chatgpt with my 2008 product dev brief (which I wrote for a dev company) - that the project could be built using agents.

I have been beta testing for the past 2 months (as this is a hobby project alongside my recently new BA role working with Salesforce, having recently relocated to USA from Australia).

What it does

What started as a repository for homeowners to collect, save and share their project journey (from before, during to finished) with friends and family.. turned into a full homeowner/provider renovation journalling tool.

The hackathon has accelerated ideas I had put on the backburner for integrating AI - where my AI feature (submission for the hackathon) is an AI-powered photo tagging feature. Its going to be the backbone of the solution to help homeowners document their renovation as it happens and turns that documentation into something they can save/search later.

Using GPT vision models to analyse renovation photos and pull out details like room type, materials, colours, finishes, fixtures and architectural features. It also then creates a detailed journal of all your renovation input for important recall later.

How we built it

I started with ChatGPT as my project planner and design architect. Where chatGPT suggested my foundations (inc starting with Vercel, v0 on Next.js and Supabase) - using Vercel's v0 agent for the early build. Which was incredible - my site was forming before my eye AND I was orchestrating features and build decisions all the way through.

However, once we progressed into polishing features.. v0 was like a bull in a china shop, it would consitently fix/change something (usually design changes as I tested) and when asked to make minor changes it broke about 10 other things. My frustration grew until I started using Codex in VS Studio... and my world changed.

Codex had the finese to fine tune/code features with fewer bugs/issues - so my project started really moving forward. I was able to create and refine more features than I expected (when planning MVP), where I now have a fully featured site in final stages of launch readiness - with FAR more of a finished product than I expected to be able to have available at first launch.

Challenges we ran into

Early days v0 was very rough in it's build - would build big silo'd features/functionality well - though for minor changes and finessing features (ie design changes, etc) idenfitified from testing - I'd constantly hit bugs and problems. Testing was getting VERY frustrating to the point where I dropped the project until I discovered Codex would be used within VS Code (which I found early 2026).

Testing was a real headache - where I had attempted to introduce playwright into v0 (not very well) to automate/simulate a lot of the test cases to iron out bugs. Again another fail for v0 - then started using Codex and it did an AMAZING job - which saved me a LOT of testing time and was able to really accelerate the build - to the point where I had run through all my features list and have been focusing on testing to be build ready.

Accomplishments that we're proud of

The fact I've been able to single-handedly build this dream site of mine (without any technical support as I'm not a coder) is something I'm quite proud of. Even if I launch and the site doesn't perform - I have always had this business idea in the back of my head for over 10 years now.. I was facing $50k+ to get it to a much lesser feature inclusive position that what I've been able to acomplish with using AI.. so it's been an amazing journey for me.. that I'm now leveaging that "orchestratoin" experience in my professional FT role as an implementation consultant using chatGPT, claude and some other Salesforce specific AI agents.

What's next for Renoview

This hackathon has been an amazing opportunity for me to bring forward some ideas that I didn't think was available - focussing on expanding the tagging concept to help the homeowner identify products they've used or paint colours, etc.. with an outcome of enriching the value of Renoview to a homeowner (to keep a subscription going) so that they have a useful journal of their renovation history.

Built With

  • codex
  • gpt5.6
  • multmodal
  • next.js
  • openai
  • postgresql
  • posthog
  • react
  • resend
  • supabase
  • tailwind
  • vercel
Share this project:

Updates