Claude Code, Grok, and the Future of App Dev with AI
from The Web Talk Show Hosted by Armando J. Perez-Carreno
Claude Code, Grok 4, and Cursor are changing how apps get built with AI. Armando J. Perez-Carreno and Mike Carlo talk about how these tools handle real code, how fast a throwaway app can come together, and why clearer prompts get you better results. In this episode: Claude Code connects to your whole repository, so it can explain a codebase you've never seen and then work on an issue, run the tests, and open a pull request. One of Mike's employees built a JSON parsing app for a client in about 45 minutes, then had the app write its own instructions for how to use it. Mike spent about an hour describing a number game from his TI-82 calculator days to Grok 4 in plain English, and it built a working version he could keep refining. The AI fixes things faster when you tell it exactly what's wrong, such as "the button showed up but the border is off." Adding images, diagrams, or Figma mockups to your prompts helps the AI get closer to what you have in mind. Guest: Mike Carlo Chapters 00:00 Welcome and how fast AI moves 01:31 Cursor and its limits with big code 03:02 How Claude Code works with your repo 05:47 A throwaway app built in 45 minutes 06:54 Building a number game with Grok 4 12:31 Why apps are becoming a commodity 13:59 Coding with AI through distractions 17:51 Directing AI and giving clear feedback 20:33 Prompting with images and diagrams 22:04 Figma MCP and how MCPs connect tools Episode page: https://webtalk.show/episodes/026-claude-code-grok-and-the-future-of-app-dev-with-ai/ Hosted by Armando J. Perez-Carreno. Learn more at https://perezcarreno.com
Embed this episode
No transcript yet.