For the last few years I've been working on VR (virtual reality) projects. Most of that work includes building EVAs (external vehicular activities) for analog environments. ISS, Mars, Moon, etc.
My work took an exciting turn with the "Summer of AI" wave that came in 2024. AI started to pop up everywhere. Local AI solutions, presentations from large companies about their products, and slowly.. research into using AI around the work place.
Starting without any prior experience, I quickly learned how to run models locally using Ollama and LangChain. These tools enabled me to generate code, answer various questions, and experiment with different models. I could also do this all offline, after downloading the models.
I have also explored using AI integrations within my personal development environment. I chose two different SDKs (software development kits)--Minecraft/Java and GTK/Rust. I wanted to see if they made using new SDKs any easier, in my main tests I treated AI as more a glorified local instance of StackOverflow. It was not a great strategy, although maybe the first path beginners might take.
I started to see the real power of AI when I was introduced into creating AI agents. Agents are made up of a graph or nodes/functions and edges/branches. Along the path of execution, a state gets updated and can be used to help determine the edges/nodes taken. In addition you can add short/long term memory and add some fine turning to parameters while the agent is running.
Part of the power is that AI can be used to help with choosing the direction of execution as well. It can be very limited levers or more open ended. I will be building more graphs and collecting experiences and perhaps sharing my own examples here.
If you're starting out with AI, I would recommend taking courses on LangGraph. it will expand your use of AI and perhaps you'll be able to integrate it into your own work or personal tasks.