A personal answer to the question I get asked most, written the week a Nobel laureate quietly changed which company owns the future.
I Picked My AI Tool, Defended It, and Almost Got Left Behind
I have a confession that is not flattering. For most of last year, I was loyal to one AI tool the way some people are loyal to a football team. I had picked it, I had learned it, and I defended it in rooms where smarter people were quietly using three others. When someone asked me which tool they should use, I gave them my answer with a little too much certainty.
Then this month happened. In the same week, the man who co-led Gemini walked out of Google to join OpenAI, and John Jumper, who shared a Nobel Prize in Chemistry for AlphaFold, left Google DeepMind for Anthropic. Two more senior researchers followed. Alphabet’s stock dropped as much as seven percent in a single day on the news.
So here is the direct answer to the question I get asked more than any other: no, you should not be loyal to one AI tool. You should be loyal to getting the best result, and you should be willing to switch the moment a better tool earns it. Loyalty to a brand in a market this fast is not commitment. It is a slow leak.
That was hard for me to admit, because admitting it meant admitting I had been doing it wrong. Let me show you what changed my mind, and what I think you should do about it.
Key Takeaways
- Brand loyalty to a single AI tool is a hidden risk, because capability is moving faster than any one company can hold it.
- When senior researchers move between AI companies, they take the next breakthrough with them, which makes talent moves a preview of where capability is heading.
- The goal is to be model-agnostic: use the best tool for each job, and stay ready to switch when a better one appears.
- You do not need to predict the winner. You need a simple habit of re-evaluating your tools every quarter.
- Documenting your prompts and workflows is what makes switching cheap instead of scary.
The Problem: We Treat AI Tools Like Marriages
Here is where I have been, and where a lot of you are right now. You found an AI tool that worked. You spent real hours learning its quirks. You built habits around it, maybe a few workflows, maybe a paid plan. And somewhere in that process, the tool stopped being a tool and started being an identity. “I’m a ChatGPT person.” “I’m a Claude person.” “I’m a Gemini person.”
I understand the pull, because I lived it. Learning a new tool is uncomfortable. Switching feels like wasting the investment you already made. And honestly, having a default answer feels good. It makes you feel like you have it figured out in a world that keeps moving the goalposts.
But I want to be honest with you, because that is the only way I know how to be useful. The comfort of loyalty is exactly what makes it dangerous. While you are defending your tool, the people who built the best tools are moving to different companies. The capability you fell in love with does not stay put. It follows the talent. And the talent is on the move right now in a way I have never seen.
I am not going to pretend this is easy. Choosing to stay flexible means choosing to stay a little uncomfortable on purpose. But what if the discomfort of switching is actually the cheapest insurance you will ever buy?
Capability Walks Out the Door
Let me give you the facts that rearranged my thinking, because they are genuinely striking.
First, the departures. Noam Shazeer, who co-led Gemini, announced he was leaving Google for OpenAI. Within days, John Jumper, the DeepMind director who led AlphaFold and shared the 2024 Nobel Prize in Chemistry, said he was leaving for Anthropic. Researchers Jonas Adler and Alexander Pritzel were reported to be heading to Anthropic as well. This is not turnover. This is the migration of the people who define what the frontier can do.
Second, the market noticed. Alphabet shares fell as much as 7.2 percent intraday on the Jumper news alone. When a single researcher leaving moves a trillion-dollar company’s stock by billions of dollars, that tells you something most of us miss: capability is not evenly distributed, and it does not stay where you left it.
Third, look at how fast the tools themselves are turning over. In just the last few weeks we have seen GPT-5.5, Google’s Gemini 3.5 Flash, and Anthropic’s Claude Fable 5 and Mythos 5, which trackers describe as a meaningful tier above the previous generation. The best tool for a given job in March is not automatically the best tool in June.
Here is the plain-language version of all that data. The AI companies are not stable planets you can orbit. They are weather systems. The brilliant people who make one platform feel magical can be at a different platform by next quarter, and the magic goes with them. If your strategy is “I picked my tool,” your strategy has an expiration date you did not choose.
What Changed for Me: From Loyalty to a Simple Rule
The shift that fixed this for me was almost embarrassingly small. I stopped asking “which tool do I use” and started asking “which tool is best for this specific job, today.”
That sounds like more work. It is actually less. Because once I let go of the need to crown one winner, I stopped wrestling with the question entirely. I gave each tool a lane. One for premium writing where voice and nuance matter. One for heavy research across long documents and images. One for anything I want to keep private or run on a tighter budget. The decision got easier, not harder, because I was matching the job to the tool instead of forcing one tool to do every job.
The second thing that changed was that I stopped treating my prompts as throwaway. I started saving the prompts and workflows that worked, in plain language, separate from any one tool. That one habit quietly removed my biggest excuse for staying loyal. The reason switching felt expensive was that all my know-how lived inside one app. Once my know-how lived in a document I owned, moving it to a new tool took an afternoon, not a month.
I want to be clear that this is not about chasing every shiny new release. It is the opposite. It is about being so calm and systematic that the hype cannot rattle you. When you have a rule and a routine, a new model launch is just a quick test, not an existential crisis.
Practical Steps: How to Stop Being Loyal Without Becoming Scattered
Here is exactly what I would do this week if I were starting over. None of this requires being technical.
1. List your real AI jobs. Write down the handful of things you actually use AI for. Writing, research, summarizing, coding, customer replies, whatever they are. Most people have four or five. You cannot match tools to jobs until the jobs are written down.
2. Give each job a best-fit tool, not a default. For each job, ask which tool produces the best result, not which one you are used to. Premium writing, large-document research, and private or budget-sensitive work often want different tools. Let them.
3. Build a portable prompt library. Save the prompts and instructions that work in a simple document you control, written so they would work in any tool. This is the single move that makes switching painless later.
4. Run a fifteen-minute monthly test. Once a month, take one important task and run it through a tool you do not normally use. You are not committing. You are just checking whether the frontier moved while you were not looking.
5. Schedule a quarterly stack review. Four times a year, sit down and ask one question: is any tool in my stack now clearly behind a better option? If yes, you switch that one lane. That is it.
6. Watch the talent, not just the launches. When you see that the people who built a platform have moved, mark that platform’s competitor as worth testing. Talent moves are a free preview of where capability is heading.
7. Tell your team the rule. If other people touch your AI work, make sure they know the principle: best tool for the job, ready to switch, document everything. Loyalty should never be the house policy.
Frequently Asked Questions
Is it really worth switching AI tools, given how long it takes to learn one?
Switching is only slow when your knowledge is trapped inside one app. If you keep a portable prompt library, moving a single task to a better tool usually takes an afternoon. The hours you save from better results pay that back fast.
How do I know when a new model is actually better and not just hyped?
Run your own real task through it and compare the output side by side with your current tool. Ignore benchmarks and headlines. The only test that matters is whether it does your specific job better.
Does using multiple AI tools get confusing or expensive?
Not if you assign each tool a clear lane by job type. A simple rule for which tool handles which task removes the confusion, and open-weight models can actually lower your overall spend.
What does the AI talent war have to do with my small business?
Talent moves signal where capability is heading next. When the people who built a top platform leave, the platforms they join are worth testing. It is free market intelligence you can use to stay ahead.
Should I just wait for one clear winner to emerge?
No. The frontier keeps moving, so a permanent winner is unlikely any time soon. Waiting means settling for yesterday’s performance. Staying flexible lets you always use whatever is best right now.
The Close: Bet on Capability, Not on Comfort
I started this year loyal to one tool and a little proud of it. I am ending this stretch with something better than pride. I have a rule, a routine, and a calm I did not have before. When the headlines scream about who left Google for Anthropic, I do not feel anxious. I feel informed.
Here is what I want you to hear, because I learned it the slightly painful way. The tool you love did not earn permanent loyalty. The people who made it great can leave, and the capability leaves with them. Your loyalty belongs to your customers, your family, your mission, your faith. It does not belong to a piece of software that could be eclipsed by next quarter.
So hold your tools loosely and your standards tightly. Use the best one for each job. Stay ready to switch. And stop letting comfort make a decision that capability should be making.
Bet on capability. It is the only thing in this market that keeps its promises.
Jonathan Mast is the founder of White Beard Strategies, where he helps tens of thousands of entrepreneurs put AI to work without losing their voice or their sanity. He is the creator of the Perfect Prompt Framework, a speaker, and a recovering one-tool loyalist who now keeps his prompts in a document he owns. He still roots for his teams. He just stopped rooting for his software.
