Developing products in the AI era.
During a round of golf the conversation turned to courses we'd love to play. We each had a list, just kept in different places. Mine lived in a chaotic Apple Note; his was a carefully ranked Google spreadsheet.
We agreed quickly that there was no simple enjoyable app that did the job well. Everything we'd tried was weighted down by ads, paywalls or features we didn't use or need. With no-/low-code tools becoming more accessible, we decided to try building something for ourselves from scratch.
How might we make it easy for golfers to save and revisit the courses they want to play?
Role: Founder & Product Lead
Team & Tools: 2 Engineers, v0, Cursor, Figma MCP
Team & Tools: 2 Engineers, v0, Cursor, Figma MCP
Duration: Ongoing
How might we make it easy for golfers to save and revisit the courses they want to play?
v0
Being able to change models yielded very different results sometime (NanoBananaPro being best)
Whilst slower, it allows for a wider context of rules and formatting.
Credits seem to go further and although sometimes smaller prompts used more than expected.
Codebase is open and allows for edits on free plan.
Lovable
Consumer grade vibe-coding at it's finest.
With the initial prompt Lovable managed a closer rendition.
Daily limit (free plan) was a very limiting factor, the context window for prompts was better.
Codebase being locked behind paywall stopped us from using it any further
Research
Research consisted of opening the topic with our friends who golfed. A group chat was set up and we led with simple questions and let them expand on what they'd like.
Everyone had their own method to keeping track, each being particularly manual and basic.
Ideas & Development
This project was entirely for ourselves, so we skipped a few steps and dived into ideas through FigJam.
We'd also be pitting vibe-coding tools against each other. I'd take v0, he'd take Lovable to develop the core flows of the app. Below is what was provided to each tool along with the initial prompt being the same.
Playlists
The idea that kicked it all off. Playlists needed to be the easiest they possible could with users being able to go through the flow in as little as 4 steps.
Playlists are also designed to be a good looking feature, we wanted to showcase beautiful courses not just random names and locations. They are also treated as a living item, we expect users to edit, refine and play around with them over time. There is also no limit to how many you can create or your entries.
Beta
Closed Invites
We gave out closed invites to the original group of friends to bug bash and provide feedback in December 2025. Already we've been inundated with bugs and feature requests.
Future
We currently have no plans to expand this to scale or open the invites. Current plans involve producing the desktop version and fixing some major bugs found. This is an ongoing personal project.
Notes:
Vibe-coding tools are brilliant…for developing rapid prototypes for validation. If you want any level of creativity, problem solving or good codebase then you still need developers and designers to step in.
I will be implementing v0/Lovable into my workflow in the future, but with a grain of salt about how they work. They allow me to spend less time fiddling with Figma Prototype settings and more time gaining real user feedback on some very early ideas.
Development
With v0 being the winning tool, we upgraded the plan and cracked on with the frontend.
I began travelling soon after and was able to work from my iPad to develop from Figma designs previously built to help styling. We managed to get the majority of the frontend tied up and developed within a month, then came the backend.
We turned to our ML friends to help us build a webscraper using Cursor . This would gather known golf courses, compile and refine them, populating the course page templates. After a weekend of scraping we had gathered 15,000 courses worldwide.
Search & Map
Our largest challenge came from integrating Mapbox and location matching to course information. Luckily our developer friends were on hand to guide us.
We now have all 15,000 courses mapped with interactive map and comprehensive search results.
User Profiles
Profiles are kept light, away from ratings, handicaps and other bragging scores. Just what you love and what you've created.
Rich Course Pages
Research told us there were too many conflicting sources of information about a golf course. We wanted to simplify the search by being a single source of truth.
We introduced an open gallery, course details (par, holes, facilities) and the ML Engineers worked on developing a AI-Generated bio for each course.
Search & Map
Our largest challenge came from integrating Mapbox and location matching to course information. Luckily our developer friends were on hand to guide us.
We now have all 15,000 courses mapped with interactive map and comprehensive search results.
User Profiles
Profiles are kept light, away from ratings, handicaps and other bragging scores. Just what you love and what you've created.
Rich Course Pages
Research told us there were too many conflicting sources of information about a golf course. We wanted to simplify the search by being a single source of truth.
We introduced an open gallery, course details (par, holes, facilities) and the ML Engineers worked on developing a AI-Generated bio for each course.
User Profiles
Profiles are kept light, away from ratings, handicaps and other bragging scores. Just what you love and what you've created.
Rich Course Pages
Research told us there were too many conflicting sources of information about a golf course. We wanted to simplify the search by being a single source of truth.
We introduced an open gallery, course details (par, holes, facilities) and the ML Engineers worked on developing a AI-Generated bio for each course.
Beta
Closed Invites
We gave out closed invites to the original group of friends to bug bash and provide feedback in December 2025. Already we've been inundated with bugs and feature requests.
Future
We currently have no plans to expand this to scale or open the invites. Current plans involve producing the desktop version and fixing some major bugs found. This is an ongoing personal project.
Notes:
Vibe-coding tools are brilliant…for developing rapid prototypes for validation. If you want any level of creativity, problem solving or good codebase then you still need developers and designers to step in.
I will be implementing v0/Lovable into my workflow in the future, but with a grain of salt about how they work. They allow me to spend less time fiddling with Figma Prototype settings and more time gaining real user feedback on some very early ideas.
