My AI Journey

Began using AI in products like Topaz Sharpen AI, Gigapixel AI, Denoise AI, etc.

Began hearing about LLMs.

Experienced AI music backing track generation in the Spark 40 guitar amplifier software.

Began using ChatGPT. This was my first exposure to LLM-based “AI” as we know it today.

It didn’t take long to realize the importance of having a conversation — of not giving up after the first response. My early use-cases were general knowledge inquiries and poetry generation. It was silly, but it was useful.

My son and I spent a lot of time talking to ChatGPT on the way to his school each day. “How long ago were dinosaurs around?… [answered]… oh, cool! Why aren’t they around anymore?”

Subscribed to Github Copilot for improved coding experience in PHPstorm and VS Code.

Everything about it was magic. It felt like it was reading my mind. I was careful, early on, to make sure it was ignoring files that typically contain sensitive information (CSV, txt, sql, etc.).

Subscribed to ChatGPT Plus.

This is the first time I started paying for AI services.

Began experimenting with LM Studio (Llama 2 chat 7B).

I was looking for a solution that could run without an internet connection.

Began experimenting with gpt4all (Llama 3 Instruct) for its local document support.

I was looking for a way to talk to an LLM with local files (PDFs mostly) as context available to the LLM.

Began using Canva AI to assist in woodworking design ideas and logo generation.

Began experimenting with Anthropic’s Claude 3.7 Sonnet.

I heard about it from a Systems Thinking YouTube channel. It seemed more polished and intellectual, so I gave it a whirl.

Began experimenting with Google’s AI Studio.

I used it to “look over my shoulder” while I worked through some design work in a new-to-me CAD application. It was incredible. I couldn’t figure out how to achieve a particular element of the design I had in my mind’s eye, and it was able to talk (literally, via microphone and speakers) me through it by watching my screen.

Subscribed to Claude after being so impressed with the user experience.

Subscribed to Cursor and began experimenting with personal projects.

It took some time to develop an appreciation for it. It took more time to trust it.

Subscribed to Google’s Gemini Pro after reading so many positive reviews on Reddit and trying it for a few days.

Really began harnessing the power of Cursor.

I developed an efficient way of getting exactly what I wanted with as few prompts as possible. I discovered the importance of keeping the context small, git-committing often, and ensuring Cursor-based changes weren’t negatively impacting the work I’d done prior.

I was starting to be able to create solutions as quickly as I could imagine them. What an incredible feeling for someone with countless software application ideas!

Subscribed to Slipbox.ai for microphone and system audio transcriptions, and realtime and post meeting support.

Being able to say “Catch me up” during a meeting is incredibly useful. “How can I contribute?” is another interesting prompt to use during a live meeting.

I wrote some automation to process the chapters and summaries after the meetings too.

Created https://www.agileadam.com/utilities as a place to share all of the micro-utilities I’d been creating.

Canceled ChatGPT Plus and Gemini Pro subscriptions.

I’m happily still paying for Github Copilot, Claude and Cursor.

Began using Anthropic’s Claude 4.0 Sonnet and Opus.