How Ordinary People Can Use Trae to Develop Smart Assistants

Today’s video content teaches everyone how to create their own smart assistant app using Trae AI with zero coding. It’s simple and easy to understand, let’s get started! (Don’t forget to check out the practical video at the end of the article!)

1. Prepare Tools

  • Open your browser and visit the Trae.AI official website.
  • Download and install the software (I have already downloaded it here).
  • Prepare a folder for project storage.

Demonstration Environment: iOS system, development tool is Xcode.

2. Initial Setup

  1. Open Trae and load the folder we prepared in advance.
  2. Open Xcode and select the same folder to load the project.
  3. Name the project: for example, “Mobile Assistant”. Keep it simple and clear.

3. Start Creating the App

  1. Go back to Trae and select a model to start working. Trae offers two models:
  • GPT-4O
  • Claude 3.5
  • We first tried the GPT-4O model, input our requirements, and let the AI start coding.
  • 4. Testing and Optimization

    • First Test Result
      : The phone only shows a home page and cannot proceed further.
    • Feedback the issue to the GPT-4O model for fixing.
    • Issue
      : The GPT-4O model repeatedly indicates that the fix is complete, but the actual test still fails.
    • Solution
      : Switch to the Claude 3.5 model to continue (Trae supports seamless model switching, which is very convenient).

    5. Continue Development

    1. Use the Claude 3.5 model to continue the task and generate code.
    2. During testing, we found:
    • The app cannot correctly understand user questions.
    • The responses are confusing, and there is a deviation in problem recognition.
  • After debugging, we found that we used OpenAI’s API, but we do not have a related account.
  • Replace it with the Deepseek API:

    Reference: For how to set up the API, please check my previous tutorials on VS and Cursor.

    • Go to Deepseek to create an API.
    • Replace the API in the Trae configuration.

    6. Final Effect

    • Prototype Functionality Completed
      : Achieved basic functions such as voice interaction and text creation.
    • Some issues remain during testing, such as repetitive answers and quick responses; the next step will be further optimization.

    Next Episode Preview We will further improve:

    • Voice tuning
    • Playing music
    • Adding more smart assistant functions

    Looking forward to your feedback: If you have any questions, feel free to leave a message in the comments, and I will reply promptly. Let’s look forward to the next content!

    Video Link: Click directly to view the actual effect demonstration.

    Respect!

    At the end of the article: The blogger is also going home for the New Year, wishing everyone a happy Spring Festival, wealth, beauty, and good health!!!

    Leave a Comment