Transcript: ‘He’s Using AI to Optimize His Life’

The transcript of AI & I with Jonny Miller is below. Watch on X or YouTube, or listen on Spotify or Apple Podcasts.

Timestamps

1. Introduction: 00:01:31
2. Dan and Jonny’s approach to running non-traditional businesses: 00:02:18
3. How Jonny uses ChatGPT to deepen his meditation practice: 00:12:04
4. Jonny uses AI to research a theory of how trauma is stored in our bodies: 00:25:44
5. Dan’s theory around how AI is changing science: 00:31:28
6. Jonny’s method to build personalized AI coaches: 00:32:35
7. How Jonny used OpenAI’s deep research to plan a move to Costa Rica: 00:47:07
8. Dan is developing an app that can predict his OCD symptoms: 00:52:50
9. AI makes the idea of a “quantified self” useful: 00:55:42
10. The future of human-AI coaching teams: 00:58:28

Transcript

Dan Shipper (00:01:30)

Jonny, welcome to the show.

Jonny Miller (00:01:31)

It's great to be here, Dan.

Dan Shipper (00:01:36)

Great to have you. So for people who don't know you, we are very good, long-time friends. You are the creator of the Nervous System Mastery course and the host of the Curious Humans podcast. You also have been an Every writer for a while. And just an incredibly interesting, fun, smart, helpful human. So I'm excited to have you on the show.

Jonny Miller (00:02:00)

Oh, that's very kind, man. Yeah, I've been getting so much inspiration from this podcast, specifically. And so I'm really excited to dive in and see what we explore.

Dan Shipper (00:02:08)

I love that. I can feel it in the notes you sent over there. They're very detailed. 

Jonny Miller (00:02:10)

Giant brain dump. Yeah, sorry for that.

Dan Shipper (00:02:12)

We have a lot to talk about. I think the place that I wanted to start actually is sort of in, I think there's a lot of overlap in the kind of business that we're building and then sort of how AI starts to fit into that. So the way that I've been talking about very recently is as a multimodal media company. So we publish writing, we publish courses, we publish podcasts, we publish videos, and we publish software. And thinking about Every as sort of this ecosystem with many different parts where we build an audience through all the stuff that all the content that we write and sometimes with the software that we make. And then we push that audience to the related products that we have. And that's how we fund the business and make great stuff. And I think you're also doing that. It may look a little bit different, but explain how you're thinking about your own business and how AI fits into that.

Jonny Miller (00:03:12)

Yeah, totally. I'd love to. So, I think this has actually been a recent conscious shift in my mind in terms of how I think about the work that I do. So, I started out as a founder, went through the startup thing, then shifted into coaching. Then my role became more like teacher and curriculum designer for Nervous System Mastery. And now I feel like I almost view my role as a steward of this emerging ecosystem. And part of that is just because there are so many different projects that I've started, I'm involved with, I'm collaborating in, and they all have a central thread. But I really enjoy bouncing between different kinds of mediums, different types of projects—some of it solo, some of it collaborative.

And I've been starting to think more about how I am almost the conductor of this orchestra or the gardener of the garden tending to different seeds and seeing where my attention is best needed. And then also how did these projects, as with Every, there's always overlap between some of the overlapping Venn diagrams. And so it's really fun for me. And obviously we can get into this, but AI is very supportive, I would say, when you're trying to do 10 things at once. There's obviously dangers to that as well, and that's something I've been trying to be conscious of in terms of what is it that I actually say no to, and how do I protect my time and not get too pulled in 1,000 different directions. But I'd say the last few months it's felt very generative, and I think it's the way that I'm wired, is to kind of be this almost constantly curious aspiring polymath-type approach to business building. And it's fun.

Dan Shipper (00:05:00)

Yeah. Me too. It sounds like we're just in exactly the same place. And I'm kind of curious how that journey has been for you. Because we both come from startups and, in the traditional startup world, you're not supposed to do things like this. First of all, you're definitely not supposed to write or make podcasts or whatever. If you're running a company, you're supposed to be doing real work.

Second of all, you must focus. You can't be doing five things at once. Because you can't do any of them well. And I think what's kind of interesting is, for me, I just want to do 50 things. I just love that. That's why I'm very curious. I want to read a lot of stuff. I want to be involved in all these different projects. I get excited about stuff and I go down rabbit holes. And I think you're pretty similar. And what's interesting to me is I think that that's just an archetype of someone who ends up being a good internet writer-type person if you have all these curiosities and you're sort of this multi-talented polymathic-type person. And a business where you can build up this sort of ecosystem is actually the right kind of business for someone like that. But it's a bit of a struggle to almost allow yourself to do that because typical business advice is: You must focus on one thing and do it really well and don't do any creative work while you're doing it. You have to manage the team and raise the money and whatever. And maybe at the beginning you built some of the product, but that's only at the beginning. What has that journey been like for you—just psychologically?

Jonny Miller (00:06:38)

Yeah, it's a great question. I'd say that, I mean, honestly, if I was like optimizing for revenue and money in the short term, that might be a better route, but I think I'm optimizing for aliveness and trusting that the more energy I have for this work, the more that that feeds into the things I create and the work that I do and the partnerships that I form. And so I'd say that some of the struggles that I have are like deciding how and where to spend my time. I think that's probably the number one thing. And I found that not giving myself too many hard deadlines has been very helpful because it allows me to move fairly fluidly, depending on where my interests are.

So I don't know, to give an example, Kate from Every sent me an email the other day being like, hey, would you like to write a guest post about AI coaching? And I was like, yeah, that actually sounds great. I'll just ditch everything I was doing and just draft this essay and there's a scenario where maybe I could have been spending my time on something else, but that was just what was exciting for me. And so that's how I've been kind of thinking about this ecosystem in terms of how it is fueling my livelihood and then also trusting that kind of tending to these projects. And for me, I'd say, I mean, maybe similar to Every, I have one big project, which is the Nervous System Mastery course. That generates a good chunk of the revenue, but then there's like the Stateshift app, there's the Calm Cards project that I'm doing. There's the podcast, there’s the writing—

Dan Shipper (00:08:30)

Well, tell us what those are just so for people who are listening, so they get a sense of the breadth of the ecosystem.

Jonny Miller (00:08:33)

Yeah, sure. So the Nervous System Mastery course is a cohort-based course. Only two a year—so, fall and spring. So a lot of spaciousness in between, which I really like the cyclicality of that as well, because it allows me to shift between different modes. There's almost writing and promotion mode, going on podcasts, then there's teaching mode, then there's reflection mode, then there's just like spaciousness for several months. And then the other ancillary side projects that are popping up are the Stateshift app, which is a kind of kind of nervous system regulation, iPhone, Android app with some AI features as well.

Dan Shipper (00:09:02)

And you're working on that with Cassius Kiani, who's also an Every EIR, so there is a lot of overlap.

Jonny Miller (00:09:06)

Exactly. Yeah. Cassius is great. He's a smart cookie, that guy. 

And Calm Cards, which is another collaboration with a company called Holstee. And that's a beautiful card deck based on nervous system regulation practices. There's the podcast, which I've been doing for five or six years now, which is free mentorship, but also where I just get to like having interesting conversations with people that I enjoy spending time with. And then there's in-person retreats, one-on-one coaching that I do sometimes, writing for people, folks like you guys. So, there's also some leadership projects that are at a very early stage, but those are for sure the main ones.

Dan Shipper (00:09:48)

In your mind how does it all fit together? Or when you say it has a unifying thread, what do you think about that? And how do you think about communicating it so other people understand it? If you do at all?

Jonny Miller (00:10:03)

Yeah, I think the unifying thread externally is promoting nervous system literacy and creating the conditions for aliveness in other people. That's probably the central thread and that looks differently depending on the project as well. But that's the central thread for most of the things that I'm working on.

Dan Shipper (00:10:28)

And how has AI changed what you're able to do or how you're doing what you're doing?

Jonny Miller (00:10:35)

Yeah. Well, I mean, one thing that comes to mind I think you said this on a podcast the other day. You feel like AI makes you excited to wake up in the morning because you can just ask really interesting questions. And I also paid for o1 pro and have Deep Research. So that's just been unbelievably fun to explore. So I'd say I just get to kind of ask questions maybe 10–15 times a day, especially when I'm outside and going for a walk. But I'd say more generally, it's definitely allowed me to become a you know 80 percent good at all of the various different hats that I wear. And I'm very conscious that I think my skills or zone of genius is really in curriculum design facilitation. I'm probably a grade-B marketer—I'm not that good in terms of business strategy as well. I give myself like B to C, so really using AI and stuff like business, admin, taxes, looking for an executive assistant. There’s so many things that I think it's just like allowing me to do the work of probably like five to six people—or more to some degree. And it's just reduced a lot of the friction that I think I would have had for my ignorance, frankly.

Dan Shipper (00:12:00)

Here's something that I just want to see—if you're willing to share. I know you're using it for Jhana practice. I want to see how you're doing that.

Jonny Miller (00:12:08)

I'd love to, yeah. Is it working? I'd say, yeah, it’s working to some degree. So, do you want me to screen share or should I just talk about it? Okay. Let's see. So, while I find this, do you want to explain what Jhanas are for people who might not be?

Dan Shipper (00:13:28)

Yeah, so if you're listening to this podcast or watching and you don't know what Jhanas are, please turn it off. I'm just kidding. They're kind of the new hot meditation state that I think a lot of the AI people—like, Nick Camerota got really into a couple years ago, and then it sort of spread to all the rationalists and Slate Star Codex had a big article about it. But basically, the idea is if you learn to direct your attention in such a way that you can attend to states of pleasure in your body, your attention works as a little bit of a feedback loop. And there are ways to reliably attend to things that feel good in your body so as to create a feedback loop, so that it sort of takes over your whole body and you achieve a Jhana, which there are different levels of and different levels of have are different feeling states. But basically a state of bliss or joy is the first, and there's all these new meditation companies, like this company called Journey, which does Jhana focus retreats. They're not a traditional lineage. They just take pragmatically from every teacher that they run into. And the idea is you can reliably access states of pleasure and joy on command—they seem to have a pretty high success rate. So I've honestly thought about going because it seems kind of fun, but yeah, that's the brief introduction to the Jhanas.

Jonny Miller (00:14:16)

Great. Awesome. So there's context from my side. So, I actually just got back from facilitating breathwork at one of the Jhanas retreats.

Dan Shipper (00:14:23)

Oh, no way. How was it?

Jonny Miller (00:14:25)

It's amazing. Yeah. I did one of their 10-day or week-long retreats last summer. And Stephen, one of the founders, has been kind of coaching me as well. So I've been using Stephen's feedback alongside ChatGPT.

Dan Shipper (00:14:27)

And did you do Jhana? What's your Jhana level?

Jonny Miller (00:14:31)

Yeah, I actually hit it on the first day.

Dan Shipper (00:14:33)

No way.

Jonny Miller (00:14:45)

Yeah. And I've been working on getting— I think, the frame that Stephen and I have is deterministic access to Jhanas one to four within 30 minutes. That's kind of the goal at the moment. And, man, it's honestly the difference between a day where I spent like 60 minutes in Jhana 4 in the morning, it's almost like a reliable indicator for like a 9 or 10 out of 10 day. It's really phenomenal.

Dan Shipper (00:15:02)

What was the thing that unhooked you or got you into it on the first day?

Jonny Miller (00:15:08)

I mean, I've done probably eight or nine extended meditation retreats before, never Jhana-specific, but I've got a fair amount of practice under my belt. And I was just really motivated, honestly. I was like, this sounds amazing. I just dove in full-steam ahead and it frankly feels like one of the most meaningful ways that I can spend my time.

Dan Shipper (00:15:35)

That's great. What is this document? What are we looking at? And for people who are listening, basically he's got open a Word document that says Jhana Expert Exploration Log. So I guess it's just a journal.

Jonny Miller (00:15:51)

Yeah. So this is essentially a journal and one of the ways that I appreciate the way that they teach Jhana practice is designing your own experiments. So instead of kind of just following a guided practice, you're more constantly trying things, trying different frames, or moving your awareness in different directions. Or, I dunno, some of the prompts are like, can I do this at a hundredth of the effort? Or one that I was playing was feeling into the humility of whatever it arises. And then I've basically been keeping a journal and this document, I just export as a PDF and then I'll upload it to this project. I've generally been using projects in ChatGPT to kind of organize, again, kind of speaking to the ecosystem thing where each project will have a different set of—obviously—project files and instructions. And so in the case of Jhana coaching, it has the journey instruction booklet that is basically kind of their recommendations for how to practice and then my own experiments and explorations. So I mean, this morning was actually a little bit of a struggle. So I could say this morning— 

Dan Shipper (00:17:23)

If you're typing into ChatGPT, I don’t see that.

Jonny Miller (00:17:12)

Oh, sorry about that. Let me just— So yeah, the instructions are relatively short here. I appreciate Rob Burbea's framing as well. So I was like, please share on the side of Rob Burbea—

Dan Shipper (00:17:24)

So, you have a project in ChatGPT—a Jhana coaching project. It has custom instructions. And it's basically asking you to reflect back what you hear. And then it also has files. So your log and then the journey manual. And so you're typing into your Jhana a coaching project: This morning, I attempted to drop into Jhana, but I felt like I was distracted and couldn't access J1.

Jonny Miller (00:18:00)

What advice do you have for me to try again later?

Dan Shipper (00:18:07)

This morning, I attempted to drop into Jhana, but felt like I was distracted and couldn't access J1. What advice do you have for me to try again later? And basically, this is with 4.5, it's going to search through all your notes, the guide, all that kind of stuff. And it's saying, it sounds like this morning's attempt to drop into J1 felt elusive because distraction was quite present. And you pulled you away from the entry point. You're seeking. This is perfectly natural. Such experiences are simply part of this subtle practice. One thing Rob Burbea might gently suggest is to start by recognizing what your experience is more than your experience this morning is not a failure or setback. How are you relating to these distractions? Was there sort of resistance, etc. Remembering from the journey instructions book, that meditation doesn't require a forcing kind of concentration, but rather something closer to gently resting in an easeful openness. I really like that. Like sunbathing on a warm day. Reflecting on your notes, a few guiding points rise and approach your next set with curiosity, effort and ease. As you noted, the aim isn't to force entry into J1 springboard practice reminder. Okay, so this is good. What do you think? And how does this compare with working with a teacher or doing it on your own? How does this make you feel?

Jonny Miller (00:19:23)

Yeah. So I think particularly the prompt around relaxing expectations— I think one of my biggest barriers is almost like trying to force my way there or having a really strong expectation to get into that state, which is actually one of the main things that blocks me. So I think just being reminded of that on a fairly regular basis is very helpful. I thought that the recommendation for starting from humility as kind of a springboard practice is often also a really nice reflection and sometimes— I didn't think this was the case this morning, but sometimes there'll be an emotion that I'm not feeling or not, or maybe not even aware of that is blocking. Stephen calls it being Jhana blocked. And it's essentially just a cue that there's something—some emotion in your system—that is like looking to be felt. And so starting with humility or, I'm sorry, can also be a nice way in as well. 

Dan Shipper (00:20:40)

What do you think I'm sorry about? Who are you apologizing to?

Jonny Miller (00:20:45)

So it's the, I'm sorry, if they have a forgiveness meditation practice that, I kind of view it through the lens of internal family systems. So sometimes there might be a part that felt rejected or maybe something happened yesterday that made me feel a sense of shame, things like that. And so just because it doesn't have to be, I'm sorry, but kind of bringing a sense of loving warmth towards it is generally really good for inviting it back in and then it will stop resisting and that will allow the Jhana juice to flow again, so to speak.

Dan Shipper (00:21:06)

Very cool. And so this is how you use it. Basically, you go in and you talk to it. Is this every day or what's the usage pattern?

Jonny Miller (00:21:15)

Yeah. So I've been trying to kind of do small journaling after the practices in the morning and then I'll add it to the Google doc and then usually on the weekends on Saturday or Sunday, I'll re-upload the PDF document to this as like an updated version and then get more meta reflections of like, okay, how do I want to approach the practice next week. And, yeah, I'd say there was a breakthrough, I'd say a week ago, 10 days ago maybe, that reinvigorated my excitement for it as well.

Dan Shipper (00:21:48)

Wow. What was it?

Jonny Miller (00:21:50)

I was able to spend probably an hour and a half in the fourth Jhana state. But it almost feels like a kind of altered state of consciousness in a way that the other three don't. And it was the first time I'd had that since the retreat in the summer. And it was just like, wow, it felt like— What's the best way to put it? I had so much clarity over almost any area of my life. And I could have like dropped in an area and I would have just had like perfect clarity around whatever that was, and combined with just a sense of a lot of ease and just deep sense of well being as well.

Dan Shipper (00:22:37)

Interesting. And did ChatGPT figure into that breakthrough at all?

Jonny Miller (00:22:55)

So I've been working with Stephen and ChatGPT for maybe three weeks. So I think the combination of the two—who's better? I mean, he's definitely more available. And I think what I liked about it as well is it's there like right after the session too. And so a way that I've more generally been using ChatGPT is: Mornings, I'll kind of do computer work, calls and things. And then I'll take our dog for a walk in the afternoon. And while I'm going for a walk, I'll just be kind of voice recording memos into ChatGPT. And Jhana practice is one of the things that I'll talk about.

Dan Shipper (00:23:25)

Interesting. So it's not Advanced Voice Mode. You're pressing the record button in ChatGPT.

Jonny Miller (00:23:30)

Exactly.

Dan Shipper (00:23:35)

Because I used to do that, but then it used to crash if you went for too long. And now I've been able to do two or three minutes without crashing, which has been good. But I don't know how long it will last. How long do you typically record into it?

Jonny Miller (00:23:48)

I typically won't do super long recordings. So I'll do maybe max two minutes and then, because you can see the text, then I'll just like maybe tweak the text if I said something wrong or if it didn't get something, but usually it's amazingly good at transcribing. And then get a response and then send another one and get a response and send another one. And then maybe summarize at the end for like takeaways of things. But that's been really fun. And I can do it like when I'm in the sunshine, I'm in the park. And that's also when I think some of the more important ideas and reflections come through as well, whether it's related to Jhana, business, personal—whatever's going on.

Dan Shipper (00:24:24)

Yeah, I love being able to think on the move and not at a desk. One of the things I've been doing recently is: In the new iOS, it's really easy to make a note and then record a voice note into the note and then it will transcribe it automatically, and then you can just press copy and paste the transcription and chat. So in the morning, I just lay on my couch and just free associate for like a couple minutes. And then I just throw it into chat. And that's another nice way to do it. And I like that too, because it means the text is in Apple Notes, so I can get it in other circumstances.

Jonny Miller (00:25:00)

Yeah, I've been trying to organize because for a while I just used chats in ChatGPT and it almost became too confusing. So I've really found that separating out different projects and specifically for the different projects in the ecosystem for Nervous System Mastery, but then also there's like one for personal reflection and there's one for a research project that I'm involved with around latches. We're moving to Costa Rica in the summer, so that's like—

Dan Shipper (00:25:32)

What are latches?

Jonny Miller (00:25:35)

Latches are— Have you heard of a guy called Mike Johnson?

Dan Shipper (00:25:40)

Maybe. It's a pretty common name. There are a couple Mike Johnsons.

Jonny Miller (00:25:44)

He's also adjacent to the Jhana community, but basically he has a fascinating theory that's building on predictive processing. It's called vasocomputation. And the TLDR of his theory is that he thinks he's found the biological mechanism for how trauma is stored in the body. But this hasn't been tested. It's still very much a theory and a thesis. And so I was using— I can maybe share the screen again here to give a specific example. But let's say, yeah, Operation Latch Hunting.

Dan Shipper (00:26:22)

So, for people who are listening, he has a project called Operation Latch Hunting. And the first chat is Vasocomputation Ultrasound Testing. What does that mean?

Jonny Miller (00:26:36)

So the TLDR of the theory is a biological mechanism of how trauma is stored in the body. And Mike's theory is that it's stored in the smooth muscle tissue using the latch mechanism. And that when someone does breathwork or a psychedelic journey, or has an emotional cathartic moment, these latches will kind of open and release.

Dan Shipper (00:26:54)

What's a latch?

Jonny Miller (00:27:00)

So the latch is how the smooth muscle tissue will kind of tense and contract, and store or buffer the emotional experience for the future, which creates a buildup of tension in the body.

Dan Shipper (00:27:10)

I went to a massage therapist the other day because I've been having back problems and I laid down and she was like, even your scalp is tense.

Jonny Miller (00:27:24)

So I've probably got some latches to do with but so the exciting thing. I'll just kind of go through some of these threads and see what we can find. But basically, I think this one was about seeing how much it knows about the latch rig mechanism and how it works. I was asking questions like, could they last months or years?

Dan Shipper (00:27:49)

Wait, can I just stop you there? You have a chat with ChatGPT. You're saying, could latches last months or years? And then it responds, lol, no, in all lowercase. Latch states are not known to last anywhere near months or years. They can be maintained for extended periods—minutes to hours in some cases, but not on those absurdly long timescales. Tell me about the personality that you've given ChatGPT here. Is that common to all of your projects?

Jonny Miller (00:28:30)

Yeah. So that's funny. So when I was doing this, I was using the Eigenrobot custom instructions. I don't know if you've seen it on Twitter, but it's all lowercase and it's like, please use Zoomer slang and give me fun abbreviations and have a sense of humor. And I enjoyed it for, I'd say like maybe a month or so. And then at some point I was like, this is kind of getting tiring. So I changed it to something that was more myself. So that’s that. But then going back to how I've been trying to use it, both trying to get a fellowship funding for Mike's research, but then some of the interesting stuff. Let's see. Not that one. Maybe it was ultrasound testing. 

So we've been thinking about, okay, how could we use an ultrasound device to actually map these latches and create a blood flow map in the body which would essentially be, okay, a map of where trauma is stored in the body. And so you could do before and after experiments, let's say, before a long breathwork retreat or before a Jhana retreat or before psychedelic-assisted therapy, etc. and so the theory was, and ChatGPT is actually incredibly helpful. Maybe not this particular thread. But in giving specific— Let's see if I can find a thread. Maybe it was this one. It's probably one of the ones in the chats, but it gave us kind of an overview of all the different blood flow mapping technology that was out there. And then it kind of gave me a list of companies to reach out to see, hey, could we get a hand on this unit and actually use it in a test. That's cool. I think this was— Yeah, so it helped to create this document, which isn't loading.

Dan Shipper (00:31:12)

So this latch is the last thing that is very interesting. I really want to know if you do an experiment with it and how it turns out I've been really into all sorts of— Okay, I've just got on my soapbox for one second. I think the idea of science being done by non-scientists is going to be more and more common. I've been doing a lot of this stuff myself, and I think a lot of the stuff makes the tools of science available much more broadly and changes, I think, what science is. So, I really want to know if this works for you. But I think one of the things that you have prepared to share or and have been working on yourself is using AI for coaching, which is a topic that is near and dear to my heart. When GPT-3 first came out and ChatGPT first came out that was one of the first things I did—started using it for therapy and look at me now. And I would love to know what you're doing and specifically. I think you're building little bots for yourself and thinking about how to teach other people how to do that. So tell us about that.

Jonny Miller (00:32:31)

Yeah, totally. So I'll just share my screen with this window here. This project kind of emerged from a workshop that I did for the Nervous System Mastery community. And so this was originally designed to be for a nervous system coach, but then I realized it could also apply to many, and I think some of the some of the insights or one of the insights I had from my own experimentation was that specifically the more context that I gave it, the more useful it was. And it seems like most people that have been trying to use it for coaching have been only really giving it context based on either the memories that it has from previous conversations. 

Or just very basic information about the problem they're having. And so what I tried to do was create a template for this thing. I call it a codex vitae, which is basically a book of life from this guy, Buster Benson, who did this back in the day. And so I basically thought about, okay, what would be the most important questions that if the AI had access to, it would be able to give me much, much, much more relevant and personalized responses. So the idea was to basically go through this list. So everything from what is my ideal weekly calendar? What are some energetic black holes, reactivity triggers, burnout warning signs, decision making principles, biomarkers, biometrics, prior experiment results. Really everything, as much self-knowledge as you have available to kind of upload into a document now that the context windows are so large, it almost doesn't seem like that's a bottleneck anymore. And so I've been obviously experimenting with this for myself. And then I think the other kind of key piece is deciding like what it is you need support with. And so this is the kind of set of custom instructions that I've been both using for myself and then adapting for some of the coaching clients that I have.

Dan Shipper (00:34:40)

And so basically this is a website, right? Where do we find this?

Jonny Miller (00:34:46)

So this I just hosted on Notion is buildyourAIcoach.com.

Dan Shipper (00:34:52)

Okay. BuildyourAIcoach.com. And it's got instructions to basically build your coach. And the first set of instructions is options for the custom instructions, the style of communication, all that kind of stuff. And it's saying you're a world-class coach. Your purpose is to provide insightful, wise reflections. You have access to my codex vitae, the document and projects file, which is what you're talking about before, which is all the context of your life. And then there's some stuff around the style of communication. Speak with direct honesty, playful, reverence and compassionate challenge. Ask potent questions. You're laconic but impactful, fewer words, deeper resonance, no wasted sentence. This is really interesting. So are you saying that this is good for everyone? Or is it just giving people jumping off points and they can sort of tweak it depending on what they like. If I like being complimented, then I could say, please compliment me.

Jonny Miller (00:35:42)

Yeah, exactly. So, this is kind of a base template. And I think there was another version where if you want to give it a name, give it a name there. And you can also share in the context of coaching—some people might have different goals, some clients are more focused on how do I become a better leader? Others are more focused on how I work with anxiety? So I think getting clear on what your specific goals are and putting that into the context piece. And then there's also a section around reflective practices, which basically give the AI coach permission to go down some of these different parts.

And so that's kind of the setup piece and then kind of going down to. So I think also cadence is somewhat interesting to talk about as well, because what I like about having a human coach is that we meet once every two weeks, there's a kind of defined time. There's also financial stakes involved as well. So I show up with a certain amount of energy and enthusiasm. And so I've added a weekly 45-minute session to my calendar where I just used to have conversations and play an experiment and then I've kind of been collecting questions or prompts that I've found useful or that I've heard other people use that have been helpful. 

So these ones are kind of more oriented towards the nervous system work—what emotions might not be avoiding help me do an energy audit, set boundaries, explore a real-time trigger, personalized supplementation template—

Dan Shipper (00:37:16)

The supplementation plan thing. Does it work?

Jonny Miller (00:37:20)

It's good, man. Yeah, it's actually surprisingly good. I mean, so I've been working with a guy called Marek, who's a fantastic functional nutritionist. And I basically compared what he recommended vs. what ChatGPT he recommended. It was very similar—and diagnosis as well.

Dan Shipper (00:37:37)

Fascinating. And what does it need to do good supplementation? What's that blood thing that Huberman promotes… one of these things that they'll send someone to your house and get labs, basically. I have that. Is that enough to do my supplementation plan?

Jonny Miller (00:38:00)

Yeah, I think a basic blood panel. Let's see if I can find it here, but I did the blood work. I did Superpower, which did the blood work. And then he recommended an organic acids test, which—

Dan Shipper (00:38:05)

Who's he?

Jonny Miller (00:38:10)

Marek, who is my functional nutritionist. And so I think this is supplement recommendation. Yeah, here we go. So let me just share this. So I was also specifically looking to improve my fertility because my wife and I are looking to start conceiving. And so this PDF was the organic acids test. It's basically saying, here's the gist and then here's the action plan: support mitochondrial function, neurotransmitter enhanced detox pathways, next steps. And yeah, checking with Marek, it seemed like it was pretty much on point.

Dan Shipper (00:38:59)

That's really interesting. And I'm curious, for this codex vitae thing, one of my experiences is I do this too on probably a smaller scale where, when I learn something about myself, I'm like, please remember this thing about me. I'm a people pleaser. So make sure that I'm not doing that or ask me. I have this thing that's been really helpful for me, which is when I'm making a decision to ask myself what I would do if I didn't feel guilty, which I think is very effective for people pleasers or at least my style. And I find that it brings it up in situations where it's not necessarily appropriate—not like it's inappropriate, but more like it probably doesn't apply or like it feels like it's trying to massage anything. Not anything, but trying to massage more things about what I'm talking about into this whole people pleaser thing, then I think is appropriate. And so have you found that? And if so, how do you work around it? Or do you just not find that?

Jonny Miller (00:40:14)

I haven’t found that. I guess maybe on a more meta point, when I get a response from ChatGPT, I'm more like looking for the nuggets of gold, as opposed to buying everything that it says. And so even if 80 percent of it is on point, then that's somewhat worth it.So to kind of maybe give a more personal example, this is the one that I created for myself, which has core personality traits.

Dan Shipper (00:40:44)

Oh, wow. We got to start from the top. So, core personality traits. This is a document. This is your codex vitae. Jonny is likely an Enneagram three with seven tendencies. What's a three?

Jonny Miller (00:40:54)

Three is the Achiever, I think, is the Enneagram name for it.

Dan Shipper (00:40:58)

And what's a seven?

Jonny Miller (00:41:00)

Seven is like an adventurer. Try 1,000 different things.

Dan Shipper (00:41:07)

So I used to be a nine, which is to see things from all angles. Like the Harmonizer, whatever, Negotiator. I can't remember. But now ChatGPT says I'm a four, which is basically the flighty creative with— It's like a one or a three or something. The Achiever I think is the wing. I think basically it's the impact of over the last couple years, I've gotten much better about just being able to know what I want in situations involving other people and not lose track of myself. It literally would be very hard for me to even know what I wanted in a lot of interpersonal situations. It would just get too stirred up and I've been much better about being able to maintain my sense of self and it turns out when I am able to maintain my sense of self, I have a lot of very strong creative desires that shift a lot. And I’ve become sort of a crazy artist, which is fine, and it was always in there. It just got swamped by this this other sensitivity that I have to other people and needs and all that kind of stuff.

Jonny Miller (00:42:34)

Yeah, totally.

Dan Shipper (00:42:35)

Okay. So Enneagram three with seven tendencies, core values, integrity, courageous, curiosity, and aliveness. Very nice. Attention and intentions and goals, preparing for fatherhood, meditating personal meditation, reliable access to Jhana one through eight, creating functional strength. Love that. What’s foiling?

Jonny Miller (00:42:56)

Oh, it's basically surfing with a hydrofoil. So it looks like a magic carpet where it goes above the water. And then—

Dan Shipper (00:43:04)

That’s sick. That's what Mark Zuckerberg was doing.

Jonny Miller (00:43:07)

Right. I don't know. Maybe.

Dan Shipper (00:43:12)

Never mind. Okay. Business: 1,700 total and some students, 1,000 paid users, 3,000. This is cool. Okay. Yeah. I have stuff like this. I have a Notion page that's basically like this.

Jonny Miller (00:43:28)

So, then there's a kind of ideal day-week architecture that's in there as well. And then this is the bit that I was going to scroll to was decision making principles. And I'll try and update this on at least a monthly basis. And this is exactly the thing you were talking about around the people-pleasing tendencies. And sometimes in a chat thread, I’ll just mention something to see if it remembers that I wrote this. And it generally seems to but I guess the question is to what degree does it really factor this in when I'm asking questions? But I'd say my experience has generally been that it has, and it's reminded me of some of these at times when it's been super helpful to do so.

Dan Shipper (00:44:09)

I have the same experience even if it sometimes overdoes it. And that is super helpful, sometimes just that one little reminder of a specific thing is enough to just totally change how you're thinking about something. And I feel like nerdy achiever people like us, we've always collected these sort of principles or mental models or whatever, but you used to have to remember them, which is a huge problem. And now you can be that kind of pack rat for interesting little tidbits or decision making principles and ChatGPT can remember it for you and you can keep the ones that seem to stick around and seem to actually work for you and get rid of the ones that don't. And I think that's so powerful.

Jonny Miller (00:44:57)

Yeah, exactly. And then, I mean, ideally over time as well. This document— So I only made this a few weeks ago and hopefully it gets better. And what I want to do is have sessions where I almost have conversations and think about, okay, how do I update this? And what are some of the questions that are most alive for me? And then have my calendar in there as well, so it's been really, really valuable and it's just been really fun as well. It's given me another, I think, reason to do a deep dive into self-exploration in recent weeks.

Dan Shipper (00:45:30)

Are you using 4.5 for this or what's your model of choice these days?

Jonny Miller (00:45:38)

I've been generally using 4.5. I think the writing, now that I've taken away the Eigenrobot instructions, I'll show you what the custom instructions I've done. Because I've thought about these quite a bit as well. This actually made a big difference to me, I'd say, in terms of the amount that I used it because it felt like it knew me more. And so let's see if I go to— Where is it? It's under share. So I've gone back and forth between Claude and ChatGPT. 

Dan Shipper (00:46:55)

Are you using Claude still?

Jonny Miller (00:46:17)

I do use Claude sometimes, I think, for writing purposes, sometimes. It gets my voice better, I’ve noticed. But let's see. So if we go customer ChatGPT. So maybe I see this one's probably better. This was a similar version to what was in the document. And then yeah, so the other thing is you can't use o1 pro on projects. So it's between o1 and 4.5. And o1 I'll use occasionally if it's like a business problem but 4.5 more for just kind of back and forth exploration and reflection.

Dan Shipper (00:47:03)

Are you using deep research? I feel like you'd be a huge nerd for deep research.

Jonny Miller (00:47:09)

I love deep research. It's unreal.

Dan Shipper (00:47:11)

Isn't it so cool?

Jonny Miller (00:47:14)

It's just so fun. I mean, to give a practical example, let's see: Costa Rica. So. We very recently made the decision to buy land and move to Costa Rica.

Dan Shipper (00:47:24)

Wait, that's awesome, but why are all your chats— Why do they have emojis? Is that a projects thing, or did you set that?

Jonny Miller (00:47:29)

I have added them. I just renamed them—renamed the chat. It's something about it. I do the same with tasks as well. It allows me to almost get an overview. I don't know. It also just looks aesthetically pleasing. So it's a weird—

Dan Shipper (00:47:49)

So, you tend to use a single chat for a long time rather than starting a new chat every time.

Jonny Miller (00:47:58)

It depends, I'd say.

Dan Shipper (00:48:00)

So you're renaming chats, even if it's a sort of more throwaway one?

Jonny Miller (00:48:06)

If it feels like a thread that I want to revisit, I'll keep it. Otherwise, I’ll just let it be or archive it.

Dan Shipper (00:48:13)

Everything looks so clean.

Jonny Miller (00:48:15)

I'm trying to find out where so I think we did a risk profile, but there was one that was— Maybe I can just search for it, basically. I was like, deep research, please figure out what are all of the logistical steps needed to move to Costa Rica. I think it was like, how to get to— Maybe, I can't find it, but it was a deep research that basically— I think it was this one, Land Purchase Guide. Yeah, so it was everything that Kelly and I needed to consider in order to move to Costa Rica. It was incredibly helpful because—

Dan Shipper (00:49:00)

Wait, can you scroll up because I want to just give people a sense for what this is. Okay, so you said, my wife and I are considering buying land in Nosara, Costa Rica. Please create a detailed brief outlining it. And it asks you for some follow-ups, which it always does, which is kind of interesting, but it's also kind of annoying because sometimes I'm like, just do it. You don't have to ask.

Jonny Miller (00:49:20)

And sometimes it's helpful, right? I mean, sometimes they say, do it. And then otherwise, in this case, it actually was helpful.

Dan Shipper (00:49:26)

Yeah. So you answered some questions and then it said, okay. So then it pre-created moving to Nosara, Costa Rica, step-by-step relocation guide, land purchase and process financing overview, buying property in Costa Rica as a foreign foreigner is very feasible. And it's just like, the most detailed step-by-step land purchase steps in Costa Rica find a reputable realtor and the right plot higher real estate estate notaries make an offer, blah, blah, blah. Some of this stuff is stuff that you would be able to find online, but it's just putting it all in a single place. And sort of tailoring it to you choosing shipping belongings from Costa Rica, what the timeline is—

Jonny Miller (00:50:03)

I wouldn't have thought of applying for residency, import taxes. There was an area on how we'd move our dog as well, which was also helpful, and how to know if we would be able to get residency there as well. It's just like everything that we needed. And this was done in 15 minutes or less. And I've been going through it, we're only a third of the way through because we're still early on in the challenge in the process, bu, bringing our dog and everything we needed to factor in. And it's just been incredibly helpful. And this was all from a one-minute prompt and so I've been doing it for that. And then also just like, I've only started adding chat threads to this, but things like: What are some of the implications of money with AI and it prepared this. Or maybe it was a different thread, but I've really been enjoying deep research. That was the thing that late night, I was like, I wonder if there's like a connection between Shamanism and Pokémon there.

Dan Shipper (00:51:26)

I would love that. That's what I want. I want to have a little section everywhere. It's like your most embarrassing ChatGPT question and people can submit it.

Jonny Miller (00:51:37)

Yeah, so this my wife was away for a couple of days and I was just pondering questions in the evening—

Dan Shipper (00:51:47)

Men need hobbies. Potential outcomes of artificial intelligence in the next five years.

Jonny Miller (00:51:53)

It’s an armchair philosopher's dream, right? It just gives you like all of these— I mean, there's half a small book here essentially. And it's also really well written as well. It's not like just AI slop. I think a lot of it is actually pretty well thought through. It's got the point where it's worth reading.

Dan Shipper (00:52:13)

One of the things that I've been doing that you might like if you're into this is it gets better with better sources. And so I built this app, I vibe coded this app that I've been calling Deep Background where I can just upload anything onto Deep Background and it turns it into a website. And so then I have a website of all my sources that is agent friendly and I can just give it to Deep Research and be like, here's a question. Here's all my notes, all my books, whatever you want, just use this site. And it works. It's pretty cool.

Jonny Miller (00:52:46)

Wow. I meant to ask you about the— You said that you were designing or vibe coding an app for your OCD and tracking tendencies. Is that still happening?

Dan Shipper (00:52:59)

Still happening. What’s funny, which maybe you'll appreciate, is how the app works— Basically, the way it works is I have a little bit of a log where I can go to the app and I take an OCD test every day. So, how are my symptoms is a standard OCD inventory called a Y box test. And then I also take a video of myself talking. I'm just talking about my day and then it sends it to this emotion embedding API. So it turns it into basically, where am I in emotional space? And the hope is I'll be able to train a model to predict from my video what my OCD score is. And then maybe, if that works, predict it a day in advance. So when is it going to get worse? When is it getting better? All that kind of stuff, which would be really fun because if you can predict it a little bit, there's all this stuff that you might be able to do treatment-wise. Maybe I just bump a little more Zoloft the day before, stuff like that, which would be kind of fun. I don't really know. So anyway, that whole thing works and I have a bunch of data. The problem is that I have not been symptomatic.

Jonny Miller (00:54:11)

Is that a problem?

Dan Shipper (00:54:13)

It's horrible.

Jonny Miller (00:54:21)

We need to trigger Dan’s OCD so he can get his app to work.

Dan Shipper (00:54:25)

I mean, that's a pretty crazy thing for me. For many years, it was a lot. It was a lot that I was symptomatic. It just has not been as much of a problem, but it has always been a little bit more episodic where it will come and go. I would say I haven't really had anything significant in three months, but it'll definitely come back and when it does, I will gather more data and then I think we'll have enough. And I'm like, they've seen the process of hiring a researcher to help me put it all together. We just have a bunch of other stuff going on Every-wise. So cash-wise and focus-wise, it's been a little bit more back burner, but it's something that we will do for sure this year. Things go well and that I think will be really, really, really cool.

Jonny Miller (00:55:16)

Yeah, it sounds amazing. And that's maybe a nice segue to something else I've been thinking about is like: AI coaches. I feel like they're decent now, certainly valuable. And I've been thinking about what are some of the innovations or things that would allow them to be even better and as good as a human. And so I'm curious what your thoughts are on that. And then I have some thoughts as well that I've been noting down.

Dan Shipper (00:55:42)

Well, I mean, the obvious one, based on what we're talking about I think a way to frame all of this is quantified self is finally useful. It used to just be you gathered a bunch of data cause you're a nerd and you're like, this is so cool. Look at my heart rate over time or whatever or look at my habits. Look at my pretty visualization that I printed into a book and published. And I was guilty. I do that too. But now we can use it to train a model to be like, well, how does X affect Y or what am I going to feel like tomorrow or whatever, or just throw it context-wise into ChatGPT and be like, what do you notice? Which I do all the time anyway with dreams. And I mean, one of the really fun things is just doing meetings and just being like, how did this go? You know? We had a big—not big—crisis, but like we had a little crisis, like a little baby crisis this morning. And I had to kind of assemble the team and be like, okay, what's, what's the status and what's going on and how do we fix it? And I was trying to figure out my strategy and I was also just interested in how I handled this? And I threw it into ChatGPT and I was like, just give me some feedback. And it helped me a lot. It’s really nice to have just like a calm, considered friendly person being like, here's what I think. And I think that that's also the case. You can think about coaching. 

There's two ways in which I can be helpful for coaching. One is as a sort of low-cost alternative for people who wouldn't ordinarily be able to have a coach or want to have a coach or a therapist which I think is super valuable and really important. And then the other one is as a sort of complement to or augmentation for a coach that almost embodies what the coach might do or become something that the coach relies on to get feedback from and basically have a relationship with, with where that thing can be available all the time because coaching as it is your coach. When someone's working with you once a week, if they really want to, and they put in the work, you can make that work. But if it's sort of like, well, I'm going to show up once a week, but you have to do all the heavy lifting. And I'm not going to think about it until next week. There's only a limited amount that you can really do with that person. But AI allows you as a coach to have much deeper insight into what's actually going on and the ability to basically have some touch points throughout the week with whoever you're working with in context. And I think that helps create more progress.

Jonny Miller (00:58:27)

Yeah. So that's something that I've been thinking about a bit as well. And I remember this was like back in maybe 2013–2014, when AlphaGo just kind of beat someone at chess. And there was this, I think Kevin Kelly maybe coined this, but it was this idea of centaurs where human plus AI teams were better than just AIs.

Dan Shipper (00:58:45)

Did he coin that? He's a beast.

Jonny Miller (00:58:50)

He is a beast. I mean, I read it from him. I assumed he coined it. But he said everything relevant to AI 10–15 years ago. But I think this idea of centaurs is actually relevant here in coaching because, for me, knowing that my clients have access to these like AI bots that I've kind of created and can not only implement the stuff that we've talked about in our session but then also continue having chats and then bring stuff to the next session. What I'd love, where I see this going would almost be a three-way chat thread where they could tag me or tag the AI, or I could see all the conversations they're having and maybe chime in if AI hallucinates or gives them advice. And I'm like, actually, no, that's a terrible idea. But it feels like such a value add. And then I'm imagining as well, once let's say you use the Hume API and maybe the Sesame-powered voice as well. And the voice crosses the uncanny valley, basically. I think that will really unlock a new mode as well. And then if you combine with what you're saying about the quantified self, if you can kind of pull in all of that data, maybe from Google or to-do list apps or APIs and the codex vitae, the deep context gets just added to it automatically without you manually having to do it. Then I think it just unlocks so much usefulness.

Dan Shipper (01:00:11)

Yeah, totally. This is actually an interesting question for me because with the OCD sort of quantified self thing, I eventually want to add other people to build models with or predict stuff. Do you have a thing that you'd want to be predicted?

Jonny Miller (01:00:30)

How likely am I to get into Jhana the next day? That's kind of cool, right? And there's actually a number of— I've been tracking factors in terms of if I sleep more than 8.5 -- 9 hours, it's more likely. If I have a less stressful day the day before, it's more likely. If I accessed it meaningfully in the previous week, it's more likely. There's actually a bunch of variables that I'm kind of tracking for myself that I think would inform. So the reliability for sure.

Dan Shipper (01:01:00)

Fascinating. Yeah, it would be fun. Because eventually what I want to be able to do is have you be able to post a bounty and be like, $10,000 if you can predict it. $20,000, if you can make it more likely. Anyway, we'll see. Interesting. 

Anything else you want to go through today? Anything else we haven't covered?

Jonny Miller (01:01:27)

No, I think that's pretty much it. There's nothing else that kind of comes to mind.

Dan Shipper (01:01:31)

Sweet. Well, Jonny, this is, this is awesome. Always a pleasure. I love what you're doing. If people are interested in your course or in your app or anything else you're doing work, where can they find you?

Jonny Miller (01:01:44)

Thanks, Dan. The best website is nsmastery.com and that's the Nervous System Mastery course. There's links to a free self assessment to the Stateshift app. And yeah, then the Curious Humans podcast is actually—they can listen to the conversation we had a few weeks ago. That’s one of the recent ones and that's just Curious Humans on most podcasts apps.

Dan Shipper (01:02:05)

That was honestly one of my favorite interviews I've ever done. People should definitely check that out. Sweet man. Well, I'm excited to hang out, I guess. Well, I'll see you. I'll see you sometime in May. So I'm excited to do that.

Jonny Miller (01:02:55)

Yeah. Epic. Epic.

Thanks to Scott Nover for editorial support.

Dan Shipper is the cofounder and CEO of Every, where he writes the Chain of Thought column and hosts the podcast AI & I. You can follow him on X at @danshipper and on LinkedIn, and Every on X at @every and on LinkedIn.

We also build AI tools for readers like you. Automate repeat writing with Spiral. Organize files automatically with Sparkle. Write something great with Lex. Deliver yourself from email with Cora.

We also do AI training, adoption, and innovation for companies. Work with us to bring AI into your organization.

Get paid for sharing Every with your friends. Join our referral program.