Skip to content
Blog & News

AI for Developers: Key Takeaways from Our Latest Tech & Consulting Session

AI is not out there to take our jobs. It is making us more efficient and making our work more meaningful. We had a learning session where our consultants dove deep into the world of AI tools for developers, featuring insightful presentations and lively discussions that showcased the potential and challenges of integrating AI into our workflows.

  • Category:

    Developing

  • Author:

    Witted Megacorp

  • Published:

The Power of RAG GPT

One of the highlights was Maks Turtiainen’s presentation on a client-specific GPT, a Retrieval Augmented Generation (RAG) application built using Azure OpenAI. Maks shared his journey in AI development, detailing how GPT has significantly impacted the operations, saving millions of euros in employee time. By combining the power of Large Language Models (LLMs) with the internal knowledge base, the client-specific GPT provides accurate and context-aware answers to complex questions, enhancing efficiency across various departments, including HR and engineering.

No engineers were harmed in making of the RAG GPT, lots of grey hair and frustration were saved. Somebody might say that it’s business process efficiency and cost savings, but at the end of the day, it’s all about people. AI tools can enhance productivity, but they should complement, not replace, human expertise.

Cursor AI: A Developer’s Companion or a Double-Edged Sword?

Jukka Miettinen’s demonstration of building a work hours tracker with Cursor AI provided a different perspective. While Cursor AI offered ease of use and rapid prototyping capabilities, Jukka highlighted several challenges, including frequent errors, unexpected code changes, and UI limitations. If you were expecting no surprises or bumps in the road on your trip to AI utopia, you’re going to get swept off your feet.

Jukka’s experience sparked a valuable discussion about the efficiency of AI tools. While they can accelerate initial development and concept demonstration, careful code review and thorough testing are crucial to avoid introducing errors. Jukka’s key takeaway was that using AI in small increments, combined with comprehensive testing, might be the most effective approach.

Slow and steady wins the race even in AI. We humans are still very much needed in the process. And that’s pretty darn good, us being humans working in IT.