The AI tool stack I actually use every day

After months of testing every AI tool that crossed my feed, here's what actually stuck. Spoiler: it's fewer tools than you'd think.

The core stack:

Claude (Anthropic) is my go-to for deep work. Writing, analysis, code generation, project planning. Cowork Mode specifically has become my default workspace for building things. The context window is massive, and the reasoning quality is noticeably better for complex tasks.

ChatGPT (OpenAI) is my quick-draw tool. Fast answers, brainstorming, generating options when I need creative breadth rather than depth. I have 4 years of conversation history here, so there's also a personal knowledge base angle that's hard to replicate.

What I tried and dropped:

A lot. I won't name them all, but the pattern was the same: cool demo, impressive first impression, then you realize the core model is doing the heavy lifting and the "tool" is just a wrapper with a nice UI. If the value is the UI and not proprietary capabilities, it's replaceable.

The key insight:

Most people don't need 15 AI subscriptions. They need 2-3 tools they actually learn to use well. The difference between someone who's "good at AI" and someone who's dabbling is depth of usage, not breadth of tools. Learn the system prompts. Learn the shortcuts. Learn what each model is actually good at versus what the marketing says.

My rule of thumb: if I can do it in Claude or ChatGPT with the right prompt, I don't need a separate tool for it. The exceptions are tools that have access to data or integrations I can't replicate with a prompt (think: tools connected to your calendar, your codebase, your email).

I'll keep updating this as my stack evolves. The AI landscape moves fast, and what's true today might be outdated in six months.