I was asked recently what I think about Claude.
The honest answer is simple: I have not used it enough to have a strong opinion yet.
I am in a season where my tools are changing because my life is changing. My work, my priorities, and the way I want to operate are shifting in real time. So my AI usage is not “pick a model and commit.” It is test, observe, adjust, repeat.
That is what led me to the actual conclusion.
There is not a single best AI model. There is no universal winner. There is only context.
The Mistake People Make When They Ask “Which One Is Best”
Most model debates treat AI like a brand comparison. That framing is backwards.
AI is not a product you choose once. It is a layer in your operating system. Your results depend less on the logo and more on the environment you build around it.
When someone asks “What is the best AI model?” what they are really asking is:
What is the best model for my workflow, my tools, my device setup, my tolerance for friction, and my goals?
Your AI Stack Is a Workflow Decision
Your stack is the combination of model, tools, integrations, device, and habits. That stack determines what “best” even means.
Here is how I think about fit, without pretending there is a global ranking:
- If you live inside Google Workspace, a model that is tightly integrated there can reduce steps and friction. That matters more than marginal quality differences.
- If your day runs through Microsoft, a model that sits inside Teams, Outlook, Excel, and Word can become the default because it meets you where the work already happens.
- If you need a flexible general-purpose system, you want something that can move from brainstorming to drafting to structured reasoning without breaking your flow.
- If your decisions depend on current events, timeliness is part of quality. In those moments, you care less about perfect phrasing and more about being anchored to what is happening now.
- If you are building in code-heavy workflows, you want structure, consistency, and the ability to sustain long sessions without drifting.
Same category. Different environment. Different answer.
The Device Matters, and Most People Ignore It
Most people compare models in isolation. That is not how usage works.
A phone is speed and convenience. It pushes you toward short prompts and quick outputs.
A laptop makes multi-step work easier. Drafts, files, tabs, and iteration become natural instead of forced.
A desktop setup changes the game again. When your workspace is designed for deep work, your AI becomes part of a system instead of a novelty you check between tasks.
AI performance is not just model-level. It is stack-level.
If your system is clean, your AI gets better. If your system is messy, your AI gets noisy.
A Simple Context Framework I Actually Use
When I am deciding what tool to use, I do not start with model names. I start with the job.
Writing and publishing: I want coherence, tone control, and the ability to refine through multiple passes.
Business operations: I want repeatable outputs, templates, and a workflow that reduces decision load.
Fast-moving information: I want speed, current context, and clear sourcing so I know what is signal versus noise.
Building software: I want structure, consistency, and the ability to stay stable through long technical threads.
The model is the engine. Your workflow is the drivetrain. Most people obsess over the engine and ignore what transfers power.
Familiarity Compounds, and It Is Underrated
Right now, I default to ChatGPT.
Not because it wins every category, but because I know it. I know how to prompt it. I know where it is strong. I know where I need to verify. I know how to turn outputs into assets I can reuse.
That familiarity compounds over time.
Switching costs are real. Not just subscriptions, but mental friction, context loss, and rebuilding your own patterns from scratch.
This is why two people can use the same model and get completely different outcomes. The difference is not the model. It is the operator and the system.
The Real Skill Is Configuration
The competitive advantage is no longer model selection.
It is configuration.
Configuration is not just choosing the right tool for the right job. It is designing how you work so the tool can actually help you. It is building a setup where AI is infrastructure, not entertainment.
In practice, configuration looks like this:
- You know when you need live context versus deep reasoning.
- You know when to automate versus when to think manually.
- You draft where it is fastest, then refine where it is cleanest.
- You store outputs so they are reusable instead of disposable.
- You build templates so your best prompts become defaults.
Most people use AI like a slot machine. Prompt in, answer out, move on.
The people who will win are the ones who treat AI like a layer in their Personal OS. They will build a stack that fits their life, compounds over time, and makes them harder to replace.