My Top 5 Ways of Using Generative AI and LLMs

If you spend any time on X, chances are you’ve seen them: screenshots of incredibly complex n8n workflows promising to automate your entire life.

These posts often try to highlight the power of AI agents, and it’s easy to feel like you’re falling behind if you’re not building similar systems.

But don’t worry: I also don’t run an army of AI agents yet. I’m also not sure if I will anytime soon.

But I do have a couple of use cases where I do actually use AI & LLMs on a (more or less) daily basis.

How LLMs Help Me Get More Done

So, what does practical AI integration look like for me?

Here’s how I leverage generative AI:

• AI-Powered Coding Assistants: Tools like GitHub Copilot and Cursor are invaluable for code completion and suggestions. While they can sometimes be too aggressive, the speed boost they provide is significant - especially features like auto-suggestion that predict your next edit.
• Gemini as a Thought Partner: I frequently use ChatGPT (though I’m currently preferring Gemini 2.5) to bounce ideas off of, research complex topics, and challenge my own assumptions. It’s about having a discussion partner - not blindly accepting AI-generated answers.
• Deep Research for Informed Content Creation: Gemini Deep Research is fantastic for gathering information quickly (ChatGPT Deep Research is similar). While the reports can be lengthy, they provide a solid starting point backed up by sources (always double-check those sources!).
• Locally Running Open LLMs for Privacy & Cost: For tasks like summarizing text or translating content, I’m increasingly turning to open large language models that I run directly on my MacBook using tools like LM Studio and Ollama. This offers privacy benefits (no sending data to third-party servers) and can be more cost-effective.
• RAG Systems for Specific Internal Projects: I recently built a RAG system with the Gemma model, Qdrant vector database, and Ollama in a livestream. This allows me to query internal documents - like financial data or past projects - and receive targeted answers powered by AI.

Of course, there’s more you can do with AI. Everybody has their own workflows and use cases. There’s also more I do with AI. I just shared some of the most practical and useful ways I use AI in my day-to-day work.