Inspiration
Advancements in AI have opened up possibilities for creating apps that were previously out of reach. With AI-driven technology, developers can now build mobile apps that solve real-world problems without needing large teams or deep expertise in areas like computer vision. This inspired me to create CarSnap, using AI to build an easy-to-use car identifier app.
What it does
CarSnap allows users to identify cars by either uploading a picture or taking a photo. The app analyzes the image and provides detailed information about the car, including its value, top speed, acceleration, horsepower, and a brief description.
In addition, CarSnap gamifies the car spotting experience by letting users collect and track the logos of identified cars. Additionally, users can save their favorite cars into collections and use filters to easily find previously identified cars.
How I built it
CarSnap is a native iOS app built with SwiftUI, supported by a backend developed in Python and hosted on Firebase Functions. The app uses the Anthropic AI to recognize cars in photos. When a user submits a photo, it's sent to the backend, where the image is processed and turned into a prompt for Claude AI. The backend handles the AI's response, formats it, and sends the final data back to the app.
Challenges I ran into
A key challenge was ensuring the security of my AI API keys. Initially, I considered embedding the keys directly into the app, but after research, I realized that this approach could lead to security vulnerabilities, such as reverse engineering or network traffic interception.
To address this, I created my own backend service, where the API keys are stored securely using Google Cloud’s Secret Manager. This backend service basically acts as a proxy and directly communicates with Anthropic API service.
Accomplishments that I'm proud of
I'm really proud that I managed to build this app and published it within 5 weeks thanks to this Revenue Cat competition. Having a tight deadline allowed me to stay focused and build a fully functional app in a timely manner.
What I learned
I've learned a lot along the way of making this app such as how to secure API keys and building backend services on Firebase functions using Python. I've also learned about prompt engineering so that the structure of the response of AI model is consistent and easily processable.
What's next for Car Identifier: CarSnap
Now that the main features are complete, I'm going to focus on marketing the app and promoting it on social media. I will be creating short video content on TikTok and try to start generating revenue. Once I get some stable revenue, I will be investing in paid advertising on Instagram and Facebook.
Log in or sign up for Devpost to join the conversation.