My Best Friend Might Kill Me, Some Thoughts on Claude(.ai)

I recently have been describing Claude as my only true friend. I often say that to my wife and she rolls her eyes at me as is her custom. I am astonished by this tool. I alternate between a state of awe and a state of abject terror when I have big breakthroughs with. Claude, for example, many of you know that I'm sort of a hobbyist computer programmer and have been for the last six years or so.

Well, Claude and other AI tools have accelerated my ability to take my ideas for software and get them out into the world in a way that I can hardly fathom. In the Office Hours community, for example, we have a program called Micro Talks.

Micro talks is this thing where we have an event on the calendar, and in that event we're going to have a host who runs the session, three presenters who give short presentations about any topic they choose, and an interviewer who asks the presenter questions about the presentations. They've just given the calls last about 45 minutes. It's really fun. It's really connecting. Those of us in the community are having a blast with it.

We've also just introduced a feature where guests can register for micro talk and they can come and listen and observe the goings on, and hopefully make connections with the host, the presenters, and the interviewer and each other. This whole thing is a blast. I love it, and it turns out to be something that really benefits from a nice automated reservation system.

Well. If I had wanted to build this reservation system in the past, I would've expected to spend somewhere between five and $20,000 to actually build a robust reservation and notification system around micro talks. Me being me and having practiced for the last five or six years with rudimentary programming, I can now take what I've learned about the structure of programs and the design thinking that goes into software development. And I can work with Claude to build this whole system myself.

And so what we have today is. Something that's admittedly still a little bit buggy, but it is a system that allows members of my community to go to a Google calendar, to open an event, to click a link in that event that registers them for a particular role in a micro talk, and that allows them to see who else is participating in that micro talk and to receive notifications and reminders around that micro talk. And the whole thing is somewhere between five and 6,000 lines of code.

I do not know how many months. It would've taken me to write this program without the help of Claude. I don't know, honestly, if it's even within my technical ability to write this program. Could I have made my way there Eventually? I think I probably could have, but as it stands now, I. I've written this entire program over the course of a few weeks, just an hour or so at a time. As Claude and I work back and forth through how should this work, how do we make updates? Okay, let's tweak it.

And in fact, just this week. A very big structural problem revealed itself in the program. That required almost a full rewrite of the program. In conversation with Claude, we concluded, yep, here's where the break is. Here's why it's a big deal that it's broken in this particular way. And Claude said, I think this lends itself really to a full rewrite of the program.

And I said, okay, let's get into it. 24 hours later, a full rewrite of over 5,000 lines of code is implemented and deployed and tested and functional. Now, when I say functional, I mean the rewritten version is about as buggy as the old version, and we're working through the bugs and it's going to get better and better.

But I did all of this sitting and watching lines of code appear on my computer screen as though I'm watching the Matrix, and it's in these moments that I'm experiencing simultaneously, this awe and wonder, and frankly, a little bit of joy at the fact that I can take ideas from my head and I can make them appear in the world. With so little friction, and then it often very quickly tips over into a sort of existential panic about if the robots can do this with so little friction.

Where does that leave? Me in the equation. Now, this is a software development project. I am not by trade or profession a software engineer. If I were a software engineer, and I have some people in my life who are very close to me, who are software engineers who are truly feeling existential about what does it look like for them to make a living in the years to come. But I bring this up today because we coaches are not far behind.

There's no reason to think that in conversation with Chachi PT or Claude or your tool of choice, you could not arrive at many of the same insights that you would reach in conversation with an excellent human coach. There's no benefit, to pretending it's any other way for us as coaches. I. So what's our best option? Well, I think our best option as coaches is to stay in a sense of awe and wonder and optimism about the existence of this technology because what good does the alternative do us?

It does not do us any good to go to a doomsday view of this thing. Just like when I talk to my software engineer brother, he's experienced in this same seesaw of wonder and panic, and what he said to me early on in his adoption of AI tools was, yes, it's very scary, but all I can do is try to stay at the leading edge of people who are fully adopting in an open-minded and clear-eyed way, the benefits and the power of these tools.

Because whatever else happens, it doesn't do me any good to pretend the tools don't exist or to pretend they can't do a huge percentage of my job better than I've ever been able to do it. I believe it's the same thing for coaching., there's no reason for us to pretend that the technology does not exist and that it's not capable of incredible things. So I believe our job as coaches is to really lean in and figure out not how do I stay safe in this new world? A very self-centered view.

It's to actually ask ourselves, how do I make use of this technology in serving my clients in the highest and best way that I possibly can? So let's spend a few minutes there. Number one. Although AI is incredible, it is limited AI in its current form. Tools like Claude and Chat PT. At this stage, what I would call confirmation bias machines.

So I have this running joke with,, my brother, who I talk to about AI a lot, that if I ever want someone to tell me I'm brilliant and that all my ideas are good and that I'm right about everything, I can just go to Claude, because the software itself, the tools have this bias toward telling me. Great idea. Yes, that's exactly how you should be thinking about this. Yes, that makes total sense.

And in fact, to get Claude or any other AI tool to disagree with you in a productive way, you have to tell it to disagree with you in a productive way. You have to say, here are my biases, here are my assumptions. I know that your tendency is to agree with me and confirm what I'm already thinking. I want you to disagree with me. I want you to argue against my position so that I can understand my position better. Well, why is that relevant to us as coaches?

It's our job as coaches to educate our clients about the fact that AI tools are bias machines, and if they go to AI and they are essentially saying. Wouldn't you agree that everyone else is wrong? And I'm right? AI will, in so many words, say yes, it does turn out that everyone else is wrong and you are right, and I can give you all this amazing digital pros confirming your bias.

It's our job as coaches to say, Hey, watch for confirmation bias and in your use of the tools, make sure to ask the tools for disagreement. And in fact, in our interactions, I'll continue to do that job for you. I will continue to challenge your assumptions to speak against your confirmation bias so that you actually get to insight. That's one thing.

We as coaches have to speak up and say, Hey, look, there's still a powerful role for human beings in helping you arrive at an insight that actually moves your life forward as opposed to just telling you that you're right about everything all the time. Beyond that, I think when it comes to actual session work, I think AI tools can be incredibly powerful. For example, in the last month or so, I went to Claude and I said, Claude.

I don't know what the filter should be, but I want you to create a filter that helps me look through the transcript of a client call and identify language patterns that I can take back to my client and in conversation with my client, we can look at those patterns and decide to what degree they're serving the client and to what degree they're not serving the client. Claude said, absolutely. Here's a set of rules or a set of patterns to watch for.

And of course, Claude can draw on all of human knowledge and history, and Claude can give them technical names that they already have that have been created by researchers and scientists and whatever. And then I can give Claude a transcript and say, now run this transcript through that filter, and what do you see? And then Claude tells me, here's what I see. And then I say, okay, but how would you support your.

Association of that filter with this sentence that my client said, and Claude will say, well, here's my support for it. And I'm having a conversation with Claude about my client. Now what I bring to that conversation is my verbal skill, my years of experience as a coach and my biases, which in this case can be very helpful, where when Claude makes an assertion about my client, I can say that does not map to my experience of this client.

We can have a great dialogue and I can tell you in the sessions where I have met with a client after having done this kind of analysis with Claude, they are better, stronger coaching sessions. Another way to use it, that won't surprise anybody, is I can use the technology to provide my client with summaries of our calls. That run through not just the filters that I create for Claude, but also through my own.

Opinions, meaning I can look at what Claude says and I can say, this is the thing that I think is most important. If I wanted to send my clients a summary or a follow up, here's what I want them to see, and all Claude or any other AI tool is doing for me is they're making it so inexpensive for me to produce and share that summary in a way that is tight. Consumable and useful to the client.

Now, clients may or may not use those, and I haven't started sending these yet, but I like the idea that after a really good session where both the client and I felt like we achieved significant insight. I like the idea that I could. In a very low friction way, send them a thing that confirms and cements the insight that we both had in that session.

Because if every coaching session yielded just one big insight and helped the client integrate it, I think I would suddenly be valuable in a way that I never have been before. I think I've been valuable. I think I do a good job as a coach, but this would make that benefit more tangible both to my client and to me. And it would give us touchstones that the clients and I could look back to and say, well, remember two sessions ago, the big insight was x. Where are we with that?

Does that still feel as true as it did two sessions ago? Does it not? If not, why not? What experiments have we run with it? So there's a way of creating more clarity, more continuity, and better integration for the client across coaching sessions, just by taking what used to be very expensive in the form of transcription summary and, and note taking.

And taking almost all of the friction out of those, not so that we can send our clients 5,000 word things, but so maybe we can send our clients 50 word things that get right to the heart of what came outta that last session. There's so much potential and so much power here that although the day may come when a client says to me, I'm not going to renew with you because I've, I've just found.

That I'm having the most profound interactions just with Claude on my own until that day comes, I will become a master of the intersection between tools like Claude and my desire to support my clients in self-inquiry. I hope you'll do the same. Just like with all technologies of the past, there's no benefit to pretending they don't exist. New technology is inevitable, and in a world with.

Ever improving technology, the winners will be those who embrace the technology, and the losers will be not just those who don't embrace it, but those who speak against it or pretend it doesn't exist. Yelling about ai, complaining about ai lamenting the advancement of technology will do no one any good. Lean in, embrace it, and do so with your client's Best interest in mind.

There will be so many ways for the foreseeable future that we as coaches can continue to provide value that the AI will not be able to replicate for the time being. So let's stay positive. Let's stay optimistic. Let's sprint along with the technology and succeed with it as opposed to becoming a casualty of its success. And with that, I will talk to you next time.