From a parking problem
to a live iPhone app.
CurbScout · A consumer product I built and run independently, using Claude Code to turn product decisions into working software.
- My role
- Independent builder
- Focus
- Product definition, implementation & release
- Evidence
- About 6 months to build · 32 cities · live iPhone app
turn useful.
Help drivers find suitable street-parking options near a destination, with navigation that supports the search. Keep the experience honest about what the data can tell us.
01
A destination is only part of the journey.
A street-parking decision involves more than finding the nearest garage. Drivers need to understand the surrounding blocks, relevant restrictions, and how to navigate those options.
I defined the product requirements for CurbScout and built the app over roughly six months. The experience brings parking information and navigation into one journey, from entering an address to exploring nearby options.
02
The difficult work was the logic behind the map.
The hardest problems involved the mathematics of suggesting parking near a particular address, representing space along the street, and coordinating navigation. These were product decisions as well as implementation problems: the recommendation has to make sense to a person approaching the destination.
I used Claude Code throughout development and a CLAUDE.md file to guide the coding agent. I defined what the app needed to do and worked through the implementation with the agent.
The product helps people evaluate where to look. Its language needs to reflect the limits of parking data and changing street conditions.
03
AI changed how much I could build myself.
I used Claude Code to turn product requirements into a working consumer app, taking responsibility for the experience and carrying it through release.
Owning both the requirements and the implementation made the feedback loop immediate. It also made the responsibility concrete: I had to decide what was useful, resolve ambiguous behavior, and get the product through release.
04
The result is a product people can try.
I built CurbScout in about six months and shipped it to the App Store. I continue to run it independently.
The app covers 32 US cities, as listed on CurbScout’s coverage page (opens in a new tab) in October 2026. The result is a complete product people can use, from entering a destination to exploring nearby street-parking options.
Product screens and public details from curbscout.app.