My AI Stack For Work & Life
I’m not an AI addict (I can stop whenever I want) but I do use it A LOT throughout the day. That shouldn’t come as a surprise. As a software engineer, I produce a lot of code. Notice I didn’t say write. Whether it’s code, emails, blog posts (just kidding 😉); as time passes, I find myself behaving more as an editor than author of (some of) my work.
The following are the AI tools I use most often. I mean for this to be a living document. The tech landscape is shifting so quickly and so violently, that I expect this will be out of date in a few month’s time. Part of the challenge in navigating today’s world is finding the balance between leveraging the best new tools vs investing in what you have already.
Coding
Claude Code
Sonnet 4.5 is an outstanding coding model and Claude Code “the harness” supercharges it. There is an entire ecosystem and workflows that have evolved around the tool. For example, I’ve created skills for Claude to leverage both Gemini and OpenAI Codex when necessary, by leveraging their respective command-line interfaces. I feel like I get the best of all words this way.
Vibe Kanban
Vibe Kanban deserves its own post. It reimagines the interface for getting work in the age of AI agents. For now, this one-minute YouTube video demonstration does an excellent job of explaining what it is and its value proposition.
https://youtu.be/TFT3KnZOOAk?si=rGPPsj3kvoH2aVY4&t=28
Warp
Warp is my terminal daily driver. It allows me to choose the best model across a variety providers based on the complexity of the task. The fact that the model machine lives within the terminal and not necessarily on the host machine is a bigger deal than it sounds. Here's a concrete example: I do most of my work on a remote machine over SSH. I often have to move files back and forth from various places on either my host machine or the remote server. Through Warp, I can have the AI agent help me move those files, and as I SSH in and out of that machine, run terminal commands on either my laptop or the remote host, all within the same conversation seamlessly.
Productivity
Claude Desktop
Claude is an incredibly useful model, but the desktop app takes it to the next level. It’s able to read/write iMessages, files and folders, and more. It can edit Apple Notes, add meetings to your iCalendar, run terminal commands, and more.
Notion
Context is king when it comes to AI. It’s no longer just about prompt engineering — the models are getting continually better at delivering value even without elaborate prompting. But, the more relevant background information you can provide the model, the better. That’s part of what makes Notion’s AI integration so powerful. The more you invest in Notion to track your life and work, the more context its AI has. It creates a powerful flywheel.
Gemini
Google cooked with Gemini 3. And it’s not just about how smart and capable the model is. The Gemini app has some very useful features like image generation and the Canvas, which makes it possible to build entire website without ever leaving the app.
ChatGPT
The main reason I still use ChatGPT is for convenience. Hitting command+space on my keyboard opens up a modal for me to immediately interact with it. Claude Desktop has a similar feature, but it’s not as good. I can’t select the model like I can with ChatGPT and Claude navigates me away to the app — I can’t interact with it directly in the modal. Also with ChatGPT, I have a shortcut on my iPhone lock screen which immediately drops me into a voice conversation. I use the voice feature constantly.