What I built instead of writing another AI tools post
Vol. I of Learn With Darin is up. Here’s what’s in it.
learn.techwithdarin.com got a full refresh this week. Vol. I is sixteen long-form guides on the AI tools most working teams actually touch, three head-to-head comparisons for when the question is “which one,” and a study companion for the new Claude Certified Architect exam.
Here’s the shape of it.
Two foundation pieces sit before any tool gets named
For first-timers. A plain-English guide for anyone who has never opened ChatGPT, Claude, or any chatbot. Thirty minutes from now they’ve done the three things that turn it from mystery to useful. No setup, no jargon, no commitment.
For practitioners. Twelve cross-cutting habits that survive a model change, a vendor switch, and whatever gets announced next month. If you’re already comfortable with one tool, this is the right place to start. Read it before any of the vendor guides.
Fourteen tool and infrastructure guides
The lineup most working teams touch:
Anthropic: Claude Code, the Claude apps, Claude Cowork
OpenAI: ChatGPT, Codex
Google: Gemini, NotebookLM, Antigravity, AI Studio
Microsoft: 365 Copilot
Mistral: Le Chat
xAI: Grok
Infrastructure: local models on your own hardware, and managed inference across Bedrock, Vertex, and Foundry
Each one is the version I wished existed before I had to learn the tool the hard way. What changes between platforms. Where the limits are. What licensing actually costs you. What works in practice versus what looks good in a demo and falls over in production.
Three head-to-heads for the “which one” questions
ChatGPT vs Claude vs Gemini for everyday writing and thinking
Claude Code vs Antigravity vs Cursor for agentic coding
Bedrock vs Vertex vs Foundry for production model hosting
These are the three decisions I get asked about most often. The comparisons are written to help you choose, not to declare a winner.
Claude Certified Architect study companion
Anthropic’s first solution-architect exam gets its own study companion. The five judgments the questions reward, the seven question patterns, and a ten-hour lab plan so you sit the exam with the muscle memory of someone who has built the thing.
What this is not
Not a course. Not a paid product. Not a 10x-your-output thread. No popup email captures, no AI-generated robot thumbnails. Just the field notes I would have wanted earlier.
If a guide saves you a week of figuring something out, that’s the point. If you find a gap or something I got wrong, the contact link goes to me directly.