Hey folks! 👋

Super excited to share something I’ve been working on for a while:

DoCoreAI – a dynamic temperature profiler for LLMs that auto-optimizes your prompts without manual tuning.

When working with AI models like GPT or Mixtral via OpenAI or Groq, I noticed a lot of devs (including me) were stuck tweaking prompts manually—changing temperature, adding random system messages, guessing what might help the output.

Fine-tuning? Too expensive and time-consuming.

Prompt engineering? Trial and error.

DoCoreAI takes your raw prompt input and dynamically adjusts:

• 🔍 Precision
• 🎨 Creativity
• 🤖 Reasoning style
• 🌡️ Temperature

All based on the context of your input — giving you more accurate, useful, and optimized responses on the fly.

It works with both OpenAI and Groq, and it’s completely open source.

This started as a side project while experimenting with customer support AI.

I wanted AI to adapt intelligently to user intent, without hardcoding tons of prompt templates or tweaking settings manually.

Now it’s evolved into something I believe can help:

• Developers
• Prompt engineers
• AI researchers
• Anyone building AI-powered tools

• Python + FastAPI
• OpenAI & Groq support
• Swagger Docs for easy testing
• Cosine similarity for reuse of past “intelligence profiles”

Check it out on GitHub: https://github.com/SajiJohnMiranda/DoCoreAI

Or explore the idea behind it: Reddit

Would love thoughts, issues, stars ⭐, and feedback from fellow builders!

Thanks Huzzler fam! 🙌

Let’s build smarter AI tools—without overengineering the hard parts.