The model wars, and why a Coast business shouldn't bet on one AI

Every week somebody asks me, "Kristina, should I be using ChatGPT or Claude? What about Gemini? My accountant says Copilot." It is a fair question. It is also the wrong starting question, and I want to explain why.

The AI landscape on any given Monday is not the AI landscape on the following Friday. There is an independent benchmark called the Arena Leaderboard that ranks AI models based on millions of blind, head-to-head comparisons made by real users. As I write this, the top three models in the text leaderboard sit within four points of each other. By the time you read this they may have reshuffled. The labs ship every few weeks. The leader changes. Then it changes again.

For a Coast business owner deciding where to put a few hundred dollars a month and a few hours of staff time, the practical question is not "which one is winning today?". It is "how do I make a decision that still makes sense in six months, when the leader has changed twice?".

Four things people lump into one

When most people say "AI", they are collapsing four very different things into a single decision.

The company is the lab that builds the underlying intelligence. Anthropic builds Claude. OpenAI builds GPT. Google builds Gemini. Microsoft does not build a model of its own at the frontier; it packages OpenAI and now Anthropic. xAI builds Grok. Meta builds Llama and gives it away.

The model is the intelligence itself, the reasoning engine. Claude Opus 4.7. GPT-5. Gemini Pro. New versions ship constantly.

The product is the consumer-facing app built on top of the model. ChatGPT is a product. Claude.ai is a product. Microsoft Copilot is a product. These are the chat windows, the subscription plans, the apps your team logs into. They are wrappers.

The system is how your business actually uses any of this. Which workflow it sits in. Which document it drafts. Which inbox it watches. Which staff member reviews the output before it goes to a customer. This is the part that lives inside your business and does not live inside any vendor's app.

Most Coast businesses I talk to are operating at the product layer and calling it AI adoption. The real work is at the system layer, and it is the only layer that is yours.

Thirty staff with personal ChatGPT logins is not an AI strategy. It is thirty handheld GPS units and no agreed navigation system.

The handheld GPS problem

Think about a handheld GPS. Useful tool. Reliable. Nobody questions its place on a boat. But nobody would design a vessel's navigation system around one.

A personal ChatGPT or Claude subscription is the handheld GPS of AI. It is genuinely useful. It builds fluency. It lets people form a view about what these tools can and cannot do. Encourage it.

But handing thirty staff a chat subscription each is not an AI adoption strategy. It is thirty people with private chat windows, no shared context, no agreed workflow, no record of what they have used the tool for, and no governance over what client data has been pasted into a third party's training pipeline. The two things, personal tool and operational system, require completely different thinking.

Why I sign up monthly

For what it is worth, I never take an annual AI subscription. Monthly, always. Twelve months is a very long time in this landscape and the flexibility is worth more than the discount. The same logic applies to your business. Lock-in is a choice. Make it deliberately.

The deeper version of this is what I call model agnosticism. It means designing your AI workflows so that the model underneath them can be swapped out without rebuilding the workflow. Your documented process describes what the AI does in your operation, not which brand of AI does it. Your policy refers to a capability ("a tool that can draft a customer reply from the last invoice and the original enquiry"), not a vendor.

Done right, it means that when the next Claude or the next GPT ships and is meaningfully better, you can move to it in an afternoon. Done wrong, it means you have built your whole business onto a particular product's quirks and you are stuck rebuilding every time the leader changes.

What this means for a Coast business this quarter

• Try more than one. If you are only ever in ChatGPT, spend an afternoon in Claude. If you have never opened Gemini, open it. You will form a much sharper sense of where each one fits.
• Write down the workflow, not the brand. "Draft a Friday recap email to the team from this week's job notes" is a workflow. Whether it runs on Claude or GPT is a detail you should be free to change.
• Keep subscriptions month-to-month while you are still figuring out your shape.
• Do not let one product become load-bearing until you have at least seen what the next two look like doing the same job.

The businesses that will be well-positioned on the Coast in two years are not the ones that picked the "right" model in 2026. They are the ones that understood the layers, kept their options open, and built systems that survive a leaderboard shuffle.

Our Blueprint maps what model-agnostic adoption looks like for your specific business, and our AI workspace setup is the layer that puts the workflow above the brand. If you want a hand, the Find your AI quick wins email is the place to start.