Forum Discussion

ai89's avatar
ai89
Copper Contributor
Feb 27, 2025

Integrating Azure OpenAI Services

Hello everyone,

I'm currently developing an AI to Human Text Converter Free platform aitohumantextconverterfree.net that aims to transform AI text into more human style engaging text. I'm exploring ways to enhance the platform's capabilities and am particularly interested in leveraging Azure's AI services.

I've been reading about the Azure OpenAI Service and its various models, such as the o-series and GPT4o, which are designed for advanced reasoning and problem solving tasks.

I have a few questions:

Model Selection: Which Azure AI models would be most suitable for refining ai generated text to make it more human style?
Integration Best Practices: Are there recommended approaches or resources for integrating Azure's ai services into existing web platforms?
Customization: Is it possible to fine-tune these models specifically for converting AI-generated text into a more natural, human-like style?

Any insights, experiences, or resources you could share would be greatly appreciated.

Thank you!

2 Replies

  • MoritzG's avatar
    MoritzG
    Iron Contributor

    Hi ai89,

    towards Model selection: I would recommend gpt-4o (or gpt-4o-mini for cost efficiency). You can deploy it as serverless/on-demand version in azure as most cost-efficient solution.

    For integration: I assume that you have some kind of authentication for your web-app. If you use Entra-ID, you can use the token and make the request directly from your webapp.
    If not, I often used an Azure Function (would be most cost efficient, alternatively e.g. App Service) as backend, which checks for correct permissions and calls the OpenAI deployment after that.
    Just be careful that you don't expose your key in your web platform.

    For customization: I would use a combination of context/system-messages and temperature/top_p settings. Lowering temperature (e.g., 0.5) can make outputs more consistent, while a slightly higher setting (e.g., 0.8) may produce more creative results. Here, keep in mind that context also counts as input tokens.

    If that is not enough, there is also the option to train models serverless.

    Hope that helps!


    Best regards,
    Moritz

    • ai89's avatar
      ai89
      Copper Contributor

      Thanks Mortiz for your response. I really appreciate it.

Resources