I have worked with AI-related tools for several years. Many of these projects have been experimental side projects or initiatives developed during hack weeks at work. In the past, this included using Dialogflow to decipher Facebook posts about speed camera locations, using OpenCV for image recognition or enabling voice control on charging towers at work.

Today, the rise of LLM technologies has opened several doors for me, allowing for relatively easy implementation of features that would have once been incredibly difficult. By closely following this rapidly evolving landscape, I look forward to gaining a deeper understanding of how to leverage AI for amazing innovations in the coming years.

So far, I have:

• Used Genkit to explore AI in AI Chef (similar to my Let's Cook website).
• Leveraged LLMs to control browser interactions for tasks like automating grocery shopping.
• Developed MCP servers at work to integrate data from external services and internal tools.
• Integrated LLMs into my Homenode home automation system for actions and long-running automation rules.
• Created prompts and Gems in Workspace Gemini to quickly collect content from my emails.

Beyond product development, I have extensively used LLMs to boost my development productivity. As an early adopter of GitHub Copilot since its beta days, I closely follow its progress and adopt new advancements as soon as they become available. I have built entire features primarily through prompting and by fine-tuning instruction files to maximize Copilot's capabilities.

One last note: Yes, I know this site is not powered by AI. For now, I suspect that most LLMs already have this website in their training data. ;)