Background
In my previous article https://www.linkedin.com/pulse/i-built-parking-app-vibe-coding-part-1-lovable-anil-jaising-cst–f6hze/ my Certified Scrum Product Owner CSPO® from Scrum Alliance students had come up with an idea for a parking app, SpotCheck with dynamic pricing for New York City. I built a prototype using Lovable.
I write about Product Managers and Product Owners. One of the hats they wear is of a scientist. Scientists experiment, so today I decided to use another AI tool Replit to build a prototype using the same product requirement document – https://conceptsandbeyond.com/wp-content/uploads/2025/07/spotcheck_prd.pdf
Replit
Replit is an AI prototyping tool that allows you to build web-based apps by describing what you want to an agent. The agent then builds the app.
My Experiment
Replit uses an Agent that one can chat with and with the help of prompts you can build an app. You can also use canned application templates to build your initial landing page or an app that can create, retrieve, update and delete (CRUD) from a database. I used the same Product Requirements Document that I had created using https://claude.ai Check out part 1 of this article for details on how to create a product requirements document. I also used the same wireframes as I had used for Lovable. My prompt:
Build a parking app for NYC parking that uses AI for dynamic parking. Use the attached product requirements document and wireframes from a competition app as a reference. You dont have to copy the user interface. Let it serve more as a requirement
This prompt built an initial version of the app
When I ran it it had runtime errors
I chatted with the Replit Agent and prompted it to debug the issue. The issue were access issues to assets that the app was looking for. The agent fixed the issue and I tried finding parking again.
This was one place that the app did something different than Lovable. Instead of giving me a map of nearby parking spots with different prices, it took me to one which was close.
The payment options were provided but not implemented. Based on what I asked the Replit agent it built the first release from the roadmap and a bit of the features of the second version. Instead of providing an entire roadmap I think every time you prompt, its important to list the features you want it to build.
My learnings:
Replit is a more robust tool that does a great job at quickly understanding requirements and building the app. The free version can definitely give you a working app before you upgrade and pay a monthly fee. Typically the speed at which it implements something is countered by the errors it tends to have when you run something it builds. The agent is smart but has to be told to fix errors that it created. Integration to Github is not as seamless or easy as Lovable. But once you integrate it once its very seamless to use.
The Github repository is public and available here if you like to play with the code further https://github.com/conceptsandbeyond/CityParkAI
Please let me know what you think? What is your experience using these tools? What do you think of AI prototyping?
If you like to be a team going through my next AI prototyping experiment join my Product Manager and Product Owner Live Workshops
CSPO workshops – https://conceptsandbeyond.com/cspo
ACSPO workshops – https://conceptsandbeyond.com/acspoworkshops
AI4PO workshops – https://www.eventbrite.com/cc/ai-for-product-owners-4377873
