Loading summary...
5 min read (86% time saved)
This video demonstrates how to rapidly build and deploy an iOS mood tracker app to TestFlight in under an hour using AI coding tools like Windsurf, emphasizing the critical role of pre-coding design in Figma and efficient AI prompting for development and debugging.
Summarize another video
Press ⌘K to quickly paste a new URL
The video introduces the concept of building iOS apps with AI, highlighting it as a significant opportunity compared to web apps due to the current complexity and lack of advanced AI coding tools for iOS. Ras, a product engineer experienced in shipping iOS apps, is brought in to demonstrate how quickly an app can be built from scratch using AI coding tools, emphasizing the unique design principles and system architecture of iOS.
Ras proposes building a simple mood tracker app with a single core function, inspired by a successful TikTok app. He stresses the importance of starting with one feature to avoid complexity, save time, and gather early user feedback. He conducts quick competitor research on the App Store, analyzing successful mood tracker apps like "How We Feel" to gather inspiration for design and functionality, focusing on simple emotion selection and journaling.
Ras explains his process of designing in Figma before coding, asserting that it ensures a clear development roadmap and allows for early feedback. He outlines the wireframes for a typical mobile app, including a welcome screen, an onboarding experience, a home screen for past entries, and a detailed entry screen. He then proceeds to design the welcome and initial mood selection screens, demonstrating how to use ChatGPT for emotion lists and incorporating design elements like gradients and button states. He highlights how Figma helps identify missing elements, like subtext, before coding, ultimately saving time.
The development phase begins using Windsurf, an AI code editor, and Expo for the project setup. Ras explains that Expo is like the Next.js for mobile development and uses an iOS simulator to preview the app. He emphasizes sending one Figma page at a time to Windsurf and providing high-level context about the app's purpose to the AI. He also advises using the chat mode in Windsurf to ask questions and confirm actions before the AI modifies the code, and using the planner mode for larger tasks. The welcome screen is generated, and then the second screen for emotion selection is built, demonstrating how the AI accurately translates the Figma designs into functional UI elements.
After building the initial screens, Ras tests the app flow in the simulator, confirming that the welcome screen transitions to the emotion selection, and the disabled button becomes active upon selection. He then implements the remaining pages, including sub-emotion selection, journal entry, home screen, and detail view with a delete function. A bug is identified where the app reverts to the welcome screen after restart instead of the home screen. Ras demonstrates how to effectively communicate with the AI to debug this issue, explaining the desired behavior (welcome screen only for first-time users, then home screen) to fix the navigation logic.
With the app functional, the next step is to submit it to TestFlight for testing. Ras uses Expo to compile the project into an iOS build and then uses EAS submit to send it to Expo's server. He navigates through the App Store Connect process, creating a new app entry and ensuring the bundle ID is correctly handled by Expo. After the build finishes (taking about 6 minutes and 47 seconds), it is submitted to TestFlight, and Ras receives an email to install and test the app on a physical iPhone, confirming its successful deployment.
Ras and the interviewer reflect on the process, confirming that building an iOS app with AI coding tools in under an hour is achievable, despite minor technical hiccups. They reiterate the value of the design process in Figma for saving time and ensuring a clear vision. Ras highlights that such a task would have taken a month with a traditional four-person team. Looking ahead, Ras suggests future features like an LLM-powered report on mood patterns for monetization and discusses marketing strategies using TikTok to reach users. He also explains the TestFlight feedback mechanism and the often challenging App Store review process, advising patience and clear communication with reviewers. The video concludes by promoting Ras's Starter Story Build Bootcamp, which covers these app development principles.
"Right now, everyone's talking about building AI web apps, but I think the real opportunity may be on the app store." — Ras
"When it comes to your first build, you really don't want to go past one feature because that adds complexity. It adds time." — Ras
"I really believe in designing things before you develop them because it ensures that you know exactly what you're going to develop." — Ras
"If you don't have this process written or or designed out, you're going to enter the code editor not fully understanding exactly where this button goes or where this button goes." — Ras
"I think a lot of people jump in too quickly into just starting to build and just starting to code when although this may feel like an annoying extra step, you're actually saving time." — Interviewer/Host
"I would treat Windsurf less as a thing that does things that you tell it to do and more of like a cooperative partner." — Ras
"Making something like this would have actually taken us a whole month with like a four person team." — Ras
"The app store review is honestly like the the worst part. So expect it to take the longest honestly out of everything." — Ras
Divyanshu AI
5 min read
Greg Isenberg
4 min read
Zed Industries
7 min read
Greg Isenberg
8 min read
Prompt Engineering
5 min read
Theo - t3․gg
6 min read
