This feels like saying “if we just removed people’s need to walk, they’d be faster.” It’s more efficient and better for one’s brain to just write. If you remove even the step of asking, much less just writing things oneself, you remove the necessary components of critical thinking. I beat a guy reliant on AI in a debate because he didn’t know how to write arguments for himself. AI is not the future.
Thanks for sharing this. Let's say a user has been using ChatGPT for a while, and has Projects set up there. Is there any way to transfer them to Claude other than cutting and pasting prompts and rerunning them?
Ruben thank you so much for making these processes easier for those of us who get overwhelmed. Question: I'm a team of 2, Claude shows Team Premium for 5 or more. Would you still recommend it for me? It looks like Premium is the only way to keep my data from training the models, yes?
I must be doing something wrong then as in projects it remembers what I have done but in cowork the active task disappears if I reboot and it doesn't remember any of the context when I ask it about continuing
Thank you for sharing this, Ruben. I really like how practical this guide is, especially the focus on showing teams the value of AI through real work instead of abstract training sessions.
I also find this fascinating from an employer brand perspective. The companies that figure out how to structure AI collaboration like this will move much faster than the ones still treating AI as an individual productivity hack.
I have noticed that just giving people access to AI doesn’t automatically change how they work.
Lots of teams “have AI” but day to day they still fall back to their usual habits because opening a blank chat and figuring out what to ask still feels like work.
Also, long term value def comes from the examples you upload! That's what teaches the system how your company writes and defines what a good output looks for you. That becomes a kind of shared memory for the team and the companies that have it will be at a huge advantage to those that don't!
humans don't change habits for tools they have to think about. that's why build the context first. then the first interaction isn't "what do I type" but just paste and done.
and examples teach Claude what good looks like for your company specifically - that’s how to avoid a generic output.
The adoption problem you're describing isn't really about Claude—it's about incentives. Individuals adopt because they directly feel the time savings (7 minutes vs 2 hours). But on a team, you've got inertia, status quo bias, and no personal consequence if someone opts out.
The real friction isn't opening a blank chat. It's that most workplaces still reward busyness over output. If your manager measures success by "does the work happen?" not "how fast can we do it?" then why would someone risk looking incompetent by experimenting with a new tool?
Projects and templates help, but the deeper play is making adoption a team-level KPI, not a suggestion. That feels corporate, but it's the only way around the organizational inertia you're actually hitting.
People think AI requires technical skills but it requires playing with it like a kid.
The problem is: nobody plays.
They try once, get a bad result, and quit.
and the fix is stupidly simple - give yourself permission to be bad at it :)
Opening a blank chat and figuring out what to type is its own job. I guess maybe it’s why a lot of people fail at AI adoption.
They'd open it, not know what to ask, close the tab, go back to doing things the old way.
Surely if that step is removed, it will work.
They just need to show up and paste their needs.
step 1 is becoming the ai guy at your job
step 2 is making sure as many people from your team is an ai guy
step 3 is only hiring ai guy or potential future ai guy
step 4 is having an ai first company
This feels like saying “if we just removed people’s need to walk, they’d be faster.” It’s more efficient and better for one’s brain to just write. If you remove even the step of asking, much less just writing things oneself, you remove the necessary components of critical thinking. I beat a guy reliant on AI in a debate because he didn’t know how to write arguments for himself. AI is not the future.
Thanks for sharing this. Let's say a user has been using ChatGPT for a while, and has Projects set up there. Is there any way to transfer them to Claude other than cutting and pasting prompts and rerunning them?
yes there is - I made a note about it here: https://substack.com/@ruben/note/c-212469253?r=5m7l8v&utm_source=notes-share-action&utm_medium=web
Hey @Ruben, paid subscriber here. Can't join slack. Could you send me the invitation link?
just sent you a message!
Thanks man 👍
you’re welcome - did you try it?
Ruben thank you so much for making these processes easier for those of us who get overwhelmed. Question: I'm a team of 2, Claude shows Team Premium for 5 or more. Would you still recommend it for me? It looks like Premium is the only way to keep my data from training the models, yes?
there is standard seat - though less available usage :)
Do you have a core folder for different projects or do you always work within the same core folder?
I don’t use Projects anymore but Cowork: https://ruben.substack.com/p/claude-cowork
I have a dedicated folder on my desktop :)
I must be doing something wrong then as in projects it remembers what I have done but in cowork the active task disappears if I reboot and it doesn't remember any of the context when I ask it about continuing
oh is the desktop app is fully up to date?
Yeah
Thanks for the guide Ruben AI Starter here… does the guide work for solo accounts as well?
sure :) you might be interested in exploring Cowork too: https://ruben.substack.com/p/claude-cowork
👏🏻
have you set it up?
I’m starting today
hoping it went well.
Hi Ruben - I am looking for those files, can't seem to find where them anywhere.
sent you a dm.
Thank you for sharing this, Ruben. I really like how practical this guide is, especially the focus on showing teams the value of AI through real work instead of abstract training sessions.
I also find this fascinating from an employer brand perspective. The companies that figure out how to structure AI collaboration like this will move much faster than the ones still treating AI as an individual productivity hack.
Really thoughtful playbook. 🩷🦩
when teams are drowning in repetitive work, AI can very much help.
it removes the friction of having no time to produce it.
He did it again 🪄 Ruben The AI Magician
thank you as always, Nelly!! :)
what’s next for me?
How should I know? You normally think of what I need before I know I need it 😅
It's crazy how Claude made all those people switch in few weeks/month
one of the fastest user shifts we've seen :)
I have noticed that just giving people access to AI doesn’t automatically change how they work.
Lots of teams “have AI” but day to day they still fall back to their usual habits because opening a blank chat and figuring out what to ask still feels like work.
Also, long term value def comes from the examples you upload! That's what teaches the system how your company writes and defines what a good output looks for you. That becomes a kind of shared memory for the team and the companies that have it will be at a huge advantage to those that don't!
humans don't change habits for tools they have to think about. that's why build the context first. then the first interaction isn't "what do I type" but just paste and done.
and examples teach Claude what good looks like for your company specifically - that’s how to avoid a generic output.
Thanks for the guide!
How would you translate this for a data science / ML team? Would you include git repos inside projects, or artifacts?
put git repos in Project so Claude has full codebase context. plus upload your data dictionaries, model cards, and experiment logs there too.
Thanks! will give this a try
sure - let me know how it goes.
The adoption problem you're describing isn't really about Claude—it's about incentives. Individuals adopt because they directly feel the time savings (7 minutes vs 2 hours). But on a team, you've got inertia, status quo bias, and no personal consequence if someone opts out.
The real friction isn't opening a blank chat. It's that most workplaces still reward busyness over output. If your manager measures success by "does the work happen?" not "how fast can we do it?" then why would someone risk looking incompetent by experimenting with a new tool?
Projects and templates help, but the deeper play is making adoption a team-level KPI, not a suggestion. That feels corporate, but it's the only way around the organizational inertia you're actually hitting.