Monthly Learning - March 2026

More talk on AI...

Roughlock Falls Trail, Lead, SD

Honestly, I have not been doing great with tracking the things I’ve learned throughout the past couple of months. So for this month’s post and because everyone seems to be talking about AI, I thought I would throw out a couple of my experiences and thoughts on the subject.

1. Coding agents are really good. I am honestly shocked at how well these coding agents can put together working code. It doesn’t always look the best, but it often works really well. I have started to enjoy coding with agents as I have done it more. The biggest concern I have for myself when using them is becoming overly reliant on them for my coding.

2. Reviewing code is less fun when working with agents. Code reviews have always been difficult for me and now that agents are able to write mountains of code super quickly, it is becoming increasingly difficult. Thankfully, tools like CodeRabbit and Opencode review exist to help check things, but it is still a time consuming and difficult task.

3. I am finding myself “talking” with AI more often. Talking through ideas and challenges with things like ChatGPT, Claude, and Gemini are helpful in working through my own thoughts. Just the other day, I was talking through our personal financial outlook as we plan out building a new house in 2027.

4. Working with image generation, even with the latest versions of it, is still frustrating… I found myself getting annoyed with Gemini’s Nano Banana when working on something I thought it would be able to do easily. This could just be prompting skill issues on my end, but it was still frustrating trying to describe exactly what I wanted and having it do something I did not want.

5. Of the three major AI companies (Google, Anthropic, and OpenAI), I’m still wrestling with which one I dislike the least. I currently pay for Warp as my primary agentic development tool, but I am thinking about adding on a subscription to one of the big three. Grok, I won’t do. I have major issues with all of the current organizations…

   1. Anthropic seems to have the best coding models, but they are kind of being jerks by cutting off or suing anyone that does not play by their rules (see recent news about Opencode not able to use Claude subscription).

   2. Google’s models seem to be falling short of the other two, at least that is what I hear.

   3. I do not trust OpenAI in any way at the moment. Examples:

      1. They just signed a major deal with the military to use their models after Anthropic said they did not agree to certain uses of their models.

      2. They are now running ads in their services.

      3. They have considered adding “adult content” to ChatGPT.

6. I have not tried anything like Claude Cowork or OpenClaw. Mostly because I don’t trust either of them not to mess something up and I have not put the time into fully understanding what they are capable of.

7. I will remain hopeful that AI will continue to improve, but not to the point where it starts taking everyone’s job. I have enjoyed using AI as a development and thought enhancement, but I am not ready to give it the steering wheel to my livelihood. God only knows where this is going to lead.

8. It is a very tough time to beginning your career right now… A lot of places are looking to AI to do the work that would normally be given to a new employee. But since AI is able to do those things so much cheaper, why do you need junior employees? Obviously, this only lasts for so long before the senior folks leave and no one has the experience to take their place. My hope is that organizations are looking towards the future to bring in new people in order for them to gain the experience necessary to effectively use AI in the future.

I hope you enjoyed this brief look into my thoughts on AI in its current state. I would love to hear how you are using AI and which of the three options you would go with.