Take a current thread, draft, or working conversation you’ve been having with one AI and use this prompt to prepare it for a second one.
Copy the response into another platform and see what changes.
Your Prompt
I want to test this discussion in another AI platform. Help me package the context clearly so another model can give me a strong response without access to our prior conversation.
Please do the following:
Summarize the main topic or problem in 3 to 5 sentences.
Identify the real goal behind this discussion. What am I actually trying to decide, write, solve, or understand?
List the most important context, assumptions, constraints, preferences, and background details that another AI would need in order to respond well.
Separate that context into two categories: Essential context: things the new AI probably needs, and Helpful context: things that may improve the response but are not critical.
Identify any terms, references, shorthand, examples, or inside assumptions that make sense in this thread but might confuse another AI without explanation.
Pull out my preferences about tone, format, audience, style, or decision-making approach if they appear in the thread.
Point out what this conversation may still be missing. What context or clarification would make the next AI’s response better?
Create a clean transfer brief I can paste into another AI tool. It should include: the objective, the relevant context, the audience or use case, any constraints or preferences, the specific question or task I want answered.
Then create a shorter version called “minimum viable context” that includes only the essentials, in case I want to test how the next AI performs with less guidance.
Finally, list 3 things I should compare when I test the second AI’s response, so I can judge whether it is actually better or just different.
Pushing Against The Grain
Everyone’s acting like switching to Claude is some kind of revelation.
Sure… it’s good. I’ve been using it more this year too. Along with still checking in on Gemini, Grok… yes, even Co-Pilot (and it's still frustratingly bad).
Claude's Co-Work is fascinating, and the prospect of installing an agent on your desktop is intriguing (though we have yet to test this out).
And yes, i do believe we should care about how friendly these platforms are willing to get with the government.
But the reaction this month has been predictably familiar.
Not because people are trying something new… but because of how quickly they’re declaring loyalty and abandonment.
We’re talking about a few million users moving around…inside a category that's approaching a billion.
But we’re already getting “ChatGPT is finished” headlines.
After one product cycle.
Meanwhile… these are $20 a month tools.
Not a $1,200 phone you’re stuck with for three years.
Not an accounting system you have to migrate your entire company onto.
You can have multiple tools side by side on your home screen, but people are still treating this like an identity statement.
Are we really going to turn our back on future improvements this soon?
ChatGPT o3 was a year ago this month. Do you remember all the image generation prompts?
It crushed 4.1, but was replaced by -5 only four months later (GPT-5 has also since been discontinued as 5.3 now sits on your app)
Are we really ready to start writing eulogies for OpenAI before 5.4 or 6.0 even shows up?
This is the pattern we humans always fall into.
We did it with Coke v Pepsi Apple vs PC iPhone vs Android QBO vs Xero Red v Blue
We don’t just use tools… we pick sides.
And the platforms love that.
Because once you pick a side, you stop evaluating. You start defending.
Establish a loyal user base. Get them to fight. Rely on human inertia and switching costs. Stop innovating. Increase prices.
The risk is not in choosing the wrong tool… but in settling for only one too early.
Instead of using the tools, we let them use us.
We should think about this more like streaming.
You don’t turn your back on Netflix forever because HBO is better this month. Or Peacock, Prime, Hulu, etc.
You keep a few constants. You check in on the others and rotate them in for a month or two.
You use what’s useful… when it’s useful.
Yes, this will be harder than opening up HBO vs Netflix tonight. This will require some work on our part to stay sharp.
Claude does feel different to me from talking to ChatGPT. One is familiar, the other not so much. One has three years of context (and a name I've given “her”, Nova), the other has a few dozen isolated threads over a year and a half.
But I've got a wife, a couple dozen friends, and over a hundred acquaintances. I don't just keep to one friend because conversations with anyone but my wife are hard and require me to re-establish context.
We have lots of friends. Maybe we need lots of AI friends too.
———
But I digress…
AI is going to move faster and do more for us than any other platform in history.
Faster improvements. Faster shifts. Faster surprises.
If you lock yourself in now… you’re saying no to what’s coming next, unless it lands on your platform.
Over $20.
