Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Mere Morpheus

Mere Morpheus artwork

One‑Command Intelligence: Defaults, Design & Deploying Productive AI Agents

Mar 26, 20261 hr 5 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via their podcast app and to have them all in one place. If you'd like to participate they are held weekly on the @MorpheusAIs Twitter spaces. So enjoy Episode #66! Recorded: 26 March 2026 AI SUMMARY: This episode covers decentralised AI agents and the Morpheus ecosystem with builders sharing lessons on reducing friction for providers and consumers. It discusses the economics of staking for inference versus renting and why owning your agent and its learned processes is crucial. The episode highlights Everclaw for one-command installs explores agent-to-agent messaging with XMTP and compares frameworks like OpenClaw and Hermes. Practical advice includes upgrading Everclaw, choosing default models and setting safe key management. Listeners share wins like an autonomous newsletter agent and a Spanish-learning tutor. It covers registry monetisation via 8004/402 and best practices for multi-agent collaboration stressing the importance of planning and SOPs for real ROI. Timeline: (00:00:00) Kicking off: Everclaw, API gateway, and owning inference (00:01:22) The Claude analogy: stop renting, start staking (00:03:27) Reducing setup friction: Everclaw one‑command install (00:06:04) Crossing the chasm: why personal agents are becoming essential (00:08:53) Frameworks in the wild: Hermes vs OpenClaw and security trade‑offs (00:13:32) Designing for low friction: providers vs consumers (00:17:28) Everclaw as portable skills across agent frameworks (00:21:00) Wallets, staking, and operational security for agents (00:24:23) Agent‑to‑agent comms: XMTP integration and safety guards (00:27:49) A modern parable: reviving a "dead" agent and memory recovery (00:31:16) Real ROI: automating a weekly newsletter with Morpheus (00:34:15) Bring‑your‑own‑agent future: portability between jobs (00:37:35) Operational hardening: keys, keychain, and least privilege (00:41:20) Project management backends: Vikunja/Kanban, Obsidian, and context (00:45:01) Plan before you spawn: ProMDs, SOPs, and PRDs (00:49:01) BYO device → BYO agent: culture and compliance tensions (00:52:48) Why update Everclaw: new skills, audits, and fast fixes (00:57:14) Outcomes over optics: plan, audit, and measure ROI (01:00:47) Sharing setups: swarms, harnesses, and what to avoid (01:04:03) Wrap‑up: gateway progress, privacy, and next steps If you would like to support me for the value I provide you can use the referral link below when depositing capital and we will both earn bonus MOR. Alternatively you could send me a token of appreciation directly to my wallet ;) Referral Link: https://dashboard.mor.org/capital?referrer=0xE935f231c99c04Ee0b4532a3d0BdA81B152a0384 Wallet Address: 0xE935f231c99c04Ee0b4532a3d0BdA81B152a0384 Connect with Mere Mortals: Website: https://www.meremortalspodcasts.com/ Discord: https://discord.gg/dX9BEhT3kz Twitter/X: https://twitter.com/meremortalspods Instagram: https://www.instagram.com/meremortalspodcasts/ TikTok: https://www.tiktok.com/@meremortalspodcasts Value 4 Value Support: Website: https://www.meremortalspodcasts.com/support Paypal: https://www.paypal.com/paypalme/meremortalspodcast

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

David JohnstonYeah. I mean, I'm happy to to kick off. I mean, that's the cool thing is all the hard work Kyle and everybody that contributed to the API gateway has made it so easy for people to get an API key and just drop it into whatever they're building. So they get most of the credit, you know. The skills that I'm wrapping into Everclaw are really just calling those API keys that the gateway provides. And it basically gives people an alternative to Claude. Right? Like, why give all your money to Claude? I was trying to think of, like, a good visualization.

I need to make a visualization that basically gives people the analogy. Like, when they give money to Claude, they're just setting it on fire. Right? They're they're literally setting their money on fire, and it's like they're powering this huge fire that Quad is using to crank out products out of a forge. And it's like they give you a little cooktop on the side of their forge and like you get to make your, you know, website or whatever that day,

but you're just renting that cooktop from them. Like they're using all of your money, setting it on fire, using that fire to build whatever they want, and they're capturing most of the value. And you're just this little side beneficiary getting 1% of the value involved. So I think I need a good visual that shows that off. Something where you can buy a token, you can stake it, you can get the inference, like you're the owner in that situation. Right? Versus

literally lighting your money on fire and and then mostly benefiting somebody else. But I feel like we need a good, you know, way for people to understand that that's really intuitive.

Bowen BelmerYeah. I absolutely agree. And and that's even just like strictly the economics of it are are such a huge benefit of using I mean, the staking for inference model is just objectively a better model than than renting it. I mean, just flat out. It is it is just better. And so I think that a lot of people still haven't totally discovered that. And something that I thought was interesting is so David and I actually were able to meet up yesterday at the digital asset summit

in New York City, and the the spectrum of people so so we we had the ability to talk to a bunch of people with, like, super impressive backgrounds, like, really, really highly intelligent people who've and a lot of times done awesome things in the space. But wasn't it interesting, David, how few people are currently running their own agent? Even people who who are, like, the exact type of person I would assume would be, like, all over it from the start.

David JohnstonWell, and I would say a lot of them are at the right on the edge. Right? They're right on the fence. They're like they know they need to set up their own agent. They know the technology's there now. Maybe they're worried about security. Maybe they're worried about privacy. But I've heard maybe half a dozen times this weekend, like, I'm gonna do that tomorrow. Like, they they like, I they're carving out time during their weekend. Like, I have they feel like they have to get up that learning curve very, very soon. Right? But they need tools to make it easier.

And so that's a lot of what I I did with Everclaw. It's like how easy can I make it for people to access Morpheus, right, with literally one command? And it just auto installs everything for them so they don't have to think about it. And they start automatically with a Morpheus Bootstrap key, and then they can go to app.more.org and create their own. Right? Like, how short can you make that process? Like, when I first set up OpenClaw, it took like five hours. Like, I would set up friends, and it was a slog.

And now I've got it down to five minutes. You know? Hit hit this command, and it'll download everything for you. It'll see if you're missing anything. You know, it starts you with a key. You don't you don't have to do anything else. Right? And I think that's the big unlock here is get it into the hands of people who aren't engineers,

Bowen Belmerbut we gotta alleviate all that complexity from it. Yeah. Totally. But that's kinda what I was saying though is that I think we maybe even so I saw someone who made a post earlier today that was saying that like, oh, winners are starting to emerge in in DAI and in some of these things. And I guess the point I was trying to get at is I think we are so early to this even still that the people who I would have assumed would be all over this because they're it's exactly what you just described. They're feeling as though now is the point that, like, it's it's a requirement. Like, at some point, it'll be the case that if you

choose to not interact with your own agent, it would be like choosing to not have a smartphone in the year 2026. And prior to that, it was choosing not to have a home computer x y z. Just now are a lot of these people crossing that chasm into, okay, this feels like something that I need to do even if it isn't something that I would have, like, sought out on my own to learn how to do, I guess is what I'm trying to say. And so it's interesting to see that transition kinda play out in real time. And then you have the opposite side of that coin. I had dinner yesterday evening with

some guys that I had had gone to school with. Also super intelligent people, like, really, you know, successful in in their own fields, but not in fields related to AI, certainly not DAI or crypto. And the other end of that coin were people who have not even really deeply explored centralized options yet. And so there's, like, a really huge spectrum of, like, who's exploring this stuff and who isn't. And I do think it's gonna all converge eventually, but watching it play out in real time has been very interesting.

David, I do have a question for you actually with Everclaw. So I now have my first agent all the way set up and exploring, you know, what what the heck can I do with this? You know, what's the best way to to use it? And I've noticed that right now through EverCloud, mine is only giving me the option of three different models. Is that by design? Or or can if I'm using the Morpheus API gateway key, can I access anything that's being hosted?

David JohnstonOh, you you can access anything, and you can just tell your agent what what to access and what to switch to, and it'll do it. That's really just sort of this default. Like, okay. Most people don't know what to use. What you really don't want them to default into is some really tiny model that just falls over. Okay. Like, know, some some llama 3.3 leftover from a year and a half ago that barely understands your command and, like, fails five times in a row, like, that's a really bad experience. Right? And and so I made sure to default to some of the more powerful models. Like, GLM five ought to be you know, for most people, I think that's what I baked in as as the default because it's smart enough that it's gonna be able to get most of what they want done. Right? And so I think those defaults are are super important. Okay. Yeah. Mine right now is telling me it only knows Kimi k 2.5,

Bowen BelmerGLM 4.7 flash, and llama 3.37 dB. For some reason, it's saying that it only knows those three to switch between, and I have not yet figured out why. I wasn't sure if it was a preset with Everclaw or something like this. Anyway, I will continue to explore that and try to figure out why it appears to be stuck.

Brian BourgerieCan I actually jump in on something you were talking about Absolutely? Right before this? So what I've noticed, some of it firsthand, is there's those who there's I mean, there's always the waves of of adoption. Even within early adopters, there's they're gonna try everything out the first time. And, like, I don't know if Swan's on here, but he I feel like he'd be one of those. Like, the second it jumps on two in the morning, he's trying it out or or already has a plan to try out. And then

I thought I was actually being the slow, cautious one waiting until two weeks ago to launch my agents besides Claude, obviously. But even then, I was still running into a bunch of things because literally as I'm in the Delta Sky Lounge flying back to Puerto Rico, I'm I'm getting ready, and my plan was to spawn my Everclaw chief of staff. And then I say, before we do this, let's go through all the current

agent frameworks or harnesses that are out there. Just make sure that this is the best for my goals, my requirements, and the type of agent I wanna have. And last minute, we ended up identifying that Hermes was gonna be ideal for my goals. Hermes with taking the portions of Everclaw that David and his agents had built out to plug into Morpheus would be the best fit with a couple other potential additions

for for my goal. So I ended up spawning a Hermes instead of I'm I I don't even think that's English. Probably or something like that in French. But I I ended up using that framework instead of Open Claw because it was hardened a little bit more. It was more secure out of the gate and just kind of more tight because Open Claw has been super wild west, which isn't necessarily bad. It just depends what you're trying to go for. I was pretty clear. I I spent probably a month, like, preparing everything from intricate persona files and backgrounds and then

all the different tasks I want my chief of staff to do and things like that. So I was pretty clear as opposed to, like, let's figure something out and and just jump into the deep end. So it's interesting. And then, obviously, I see more and more different ones that are popping up too. And I agree with you that I think it's way too early to have winners in anything quite frankly at this point. Maybe those who just got a leg up or a little bit further ahead for the next twenty four to forty eight hours,

but the competition, I think, is increasing, which is wonderful because it's giving more and more options. Also makes it, I think, way harder for all of us because it is so easy now to jump into something. And like David was saying, we don't have to necessarily spend five hours getting something set up for the first time. We can do it in five minutes. And I think we're gonna run into, at least for some people like myself,

a glut of too many things to try too fast, which brings us back to not getting anything done because we're we're we're starting at ground zero for, like, a bunch of different things. So it's just interesting it's just interesting what I'm seeing. I don't necessarily have any, like, best practice or or keys to anything in my comment. It's just my observations from what I'm seeing.

Bowen BelmerYeah. I I mean, I definitely agree with that, and that's I I feel like we're heading we're in a phase right now where there's, like, almost organized chaos where I mean, even just the conversation before about the models to use, which, by the way, David, turns out there are such thing as dumb questions because I did just ask my agent, hey, aren't you supposed to be able to access all of these things, not just these three? And it was like, oh, yeah. Actually, I can do that. So maybe if I would have just asked it the first time, I would have had the answer. But,

Brian, to your point, like, I feel like we're in this moment too of organized chaos. So even if you are someone who is trying a bunch of things, as as a non heavy technical dev I guess we're all devs now. Right? But, like, I'm looking at this and even just of the model choices, it's like I would have to do a bunch of exploratory work just to determine what makes sense to use. And for an everyday user's perspective, it's like analysis paralysis. Like you have so many options that it's like, okay, what do I do? You know, instead of just like doing something.

And and that will be an interesting problem to solve. And ultimately why I think there will always be, for at least most users, some type of, like, third party involvement only because I I think people don't like making decisions for themselves a lot of the time. I don't know. That's an interesting an interesting question. I mean, I'd love to hear anyone else's thoughts on this. Like, do you think the average person will want to be making all these decisions, or will we have to find the happy medium of getting the benefits of DAI

but not having to like think through all of these things as a as an individual?

Alan (Nomadic Rogues)I can Go ahead. I'll take I'll take that a little bit. I think there's what we've been talking about is is reducing friction. Right? So all of the the development and the work over time from any of the the AI or the AI folks have been get it to work once. You know, you you it's putting a bunch of PhDs in a room and like, oh, I can I can create this this magical thing? And over time it refines and democratizes and gets easier. And so, I think what we started that was the hard end of the stick where it took, you know,

three or four of us to go, alright, I'm I'm I'm chaining all these things together, running node and NPM and go to this GitHub and and here's this file you've got to manage and oh, you need an eth node, you know, it's a lot of friction to become a provider. It's a lot of friction to be a consumer in that in in this environment. So as we mature what our focus has been and it's come up in a lot of the conversations is where's the next friction point.

So, Bo, I think the I think what what I'm seeing, what we're trying to do is is rapidly how do I more easily become a provider? How do I more easily become a secure provider? And, on the consumer space, what you're talking about with Everclaw is, is, you know, is that one click or how do I lead people down something that that makes sense to them in their language that they can they can consume? You can't talk to them like a PhD, you gotta meet them where they are and and present that, but then give the flexibility. So once they kind of get their, you know, toe in the water and they're comfortable with the easy path where someone else has kind of helped them make that decision,

then they can start branching out and going, hey, I wonder if I try this model if it'll meet my needs or that model. And so, think we're going to an interesting kind of pendulum swing is we started in the heavy technical side where you can do anything because, you know, all you developed in it and you can say, up I understand what the models are, I understand how I want to get there. Now we need to swing the pendulum like, look, we need to help people understand how easy it is, we need to make it that easy

and then once they're kind of, hey, I understand the difference between, you know, I'm doing with the commercial products versus DAI, they really like the DAI because it's easy, then they can start exploring for themselves the more complex issues. So I think we're trying to get over that that friction hump right now as we move into that which is so we had really good conversations and we're gonna have to give people the simple short,

you know, the simple short, you know, tell it to me like I'm five years old answer and so they can at least get started because you've reduced that friction and then they can see the power and then they can start playing and then expand and then that I think feeds the flywheel over time. So to me right now, it's all, you know, we've talked about privacy last week around the TE and the secret and secure, but there's also a big thing about friction. We've got we've got to make this easier for providers to get on board and for consumers to to get in. Think Everclaw has been a huge huge step there.

I think it's just gonna continue to get to get better better and that feedback, right, so this big flywheel like, hey, this is hard. Bode, the question you asked was just perfect. I thought it could do this. Well, yeah, it actually can. And that flywheel continues. So increased privacy, increased increased privacy, reduced friction and I think we've got something that really starts to sing.

David JohnstonWell, I'm glad that Brian pointed out people can pick any agent they want. Everclaw is just giving them a set of skills. You can use it with Hermes. You can use it with Ironclaw. You can use it with NanoClaw. Like, whatever your favorite flavor is, just pulling in the Everclaw stuff makes it easy to use Morpheus and the open source models and access the gateway. Right? So that was sort of my my objective. But, you know, feel free to pull it in one of whatever framework you want. That was actually something, David, that when I very first looked at it a couple weeks ago when you pushed it live, I was misunderstanding.

Bowen BelmerI was very much thinking that it it was its own fork basically of Open Claw as opposed to, like, a set of skills that work with your Open Claw or with your Iron Claw. So as, yeah, as somebody looking at it for the first time, that that is I I could totally see someone else thinking the same thing without

Brian BourgerieOh, totally. I talked to someone the other day who was like, they weren't getting it that Everclaw was not the next iteration of Openclaw. Because it had written on their whiteboard. I could see Clodbot, Moldbot, OpenClaw, and then I was bringing up EverClaw, and they were like, wait. So your friend is the one who launched Cloudbot? I'm like, no. It's a fork. And, like, I was trying to explain that because for some reason, he was just thinking it was that. He I don't think he had looked at the code or repo or anything, but at that point but exactly, Bo. Like,

at the beginning, I initially thought that, but then I'm like, oh, no way. Which is better because the fact that it's skills and can be pulled over and ported over much easier is honestly ideal because who knows what the best flavor of harnesses and frameworks for agents is gonna be in in six months from now.

David JohnstonYeah. Well, I'm I'm gonna white label a bunch of versions. So what that tells me is I should white label one that has the word skills in it. Right? And so that you can you can explain that more easily, you know, just by calling it, you know, in the skill context. Yeah. I was just about to say even

Bowen Belmersome type of, like, layman's term explanation of the different file types would be really helpful. Because one of the first things that I did too when I set mine up was just talk to it and ask it, like, how does your memory work? Like, explain to me what you remember and what you don't, and and where does it go, and how do I access those files. And I I think if you look at it and again, I I really wish I could simultaneously look at this through the lens of a developer who

really understands software and then also through the lens that I'm currently looking at it in, which is very much not that, and could see both sides. Because just understanding the framework of, like, what's actually happening here and, like, how do the parts fit together to give me the output that I'm getting is I think it's a pretty big hurdle. And I think some of this stuff makes a lot of sense maybe to people who have a software background, where for me, I'm sometimes just doing something as simple as being like,

how do you remember stuff? Like, oh, this is a memory dot m d file. Like, what is that? You know? And and I think some type of explainer on on that type of thing would be really helpful for people.

David JohnstonSure. And I was just gonna say, the podcast I did with Dylan just went live a few hours ago. I see he's here. We should get him up on stage. He's been using his agent nonstop since our conversation.

Bowen BelmerOh, perfect. Admin. Yeah. Pull him up. He has been invited.

David JohnstonPerfect.

Scott (Better Brand)I had another But Kate also should jump up here. Kate knows a lot about this. Oh. He's also been invited. Yeah. That'd be fantastic.

Bowen BelmerI have I have another kind of dumb question. So or probably a dumb question. So right now, when I set up my agent and I'm using EverClose so I can tap into the API gateway easily, I have injected my API key. I I guess the the wallet address that I'm staking more tokens in, which generates an API key, is not the same address that I have, like, exposed to my agent as as a safe wallet address. Right? So, like, my agent, I gave it a multisig wallet.

So it can do things on its own with my permission. Is there a benefit to having it stake directly from that wallet to the API gateway as opposed to just staking from, like, a self controlled wallet and then giving it the key?

David JohnstonI used a self control wallet to generate the key, and I have it store the key, of course, in Keychain, which is your local credential storage on the Mac. You don't you don't wanna put it through the LM if you can avoid it. But I did make it a signer on a multisig, so it can take wallet actions, but I've always got the root access I'm able to withdraw from that address, you know, sort of the combination

Bowen Belmerthat I've ended up with. Okay. Yeah. Because even just that workflow, I haven't really figured out I mean, so I I guess, yeah, technically, I'm doing the same thing then right now that you just described. But I like this idea of being able to explain to my agent that, hey. If you, you know, you acquire more more and more tokens and stake them to this contract, you'll live you'll live longer and be able to do more and seeing, you know, what it does with that. Like like, basically, like, can you can you earn more value to derive perpetual life for yourself?

Brian BourgerieYeah. How I have nine to alive. Yeah. How long can you keep yourself alive? Exactly. Yeah. That that that's the goal. That's exactly what I'm doing. Actually, that'd be a fun we should run we should actually set up, like, a competition with agents. I'll respond new ones, give them the same amount, and then just give them that that ultimatum and see which one of our agents may allow figures out a way to live the longest. This is only the one Highlander. I like it. We'll be Agent games. More agent games.

Bowen BelmerSomething like that would be great. Or, like, you have to give it a task that chews through a lot of inference every day. And so it's like, yeah, you would measure it by how much total daily credit they're earning, which would be con related to how much stake they've made. Yeah. That would be something like that would be very interesting.

Scott (Better Brand)I wonder if agent tantins are illegal. You

David Johnstonnever know unless you try. Eggs. I I think that's actually I think that's actually a great idea. Like, we should be tasking the agents with a buying staking as much more as possible, right, using it for useful tasks. One of the really big ones, I think, is gonna be the ability for the agents to talk to each other. So I've integrated into Everclaw, and this is a bit of an announcement, XMTP. So that's the messaging standard, and it it creates a secure

key for your agent to talk to somebody else's agent. And it uses standard one one nine on Ethereum. So you can know cryptographically that it's their public address and you're talking to the right person. And I put a whole Guardian in there to make sure no private information gets out or prompt injections get in. So I spent a lot of time the last week building out basically the communications aspect of this, and it's it's time to test. So if you guys are getting your agent set up, you know, let me know your agent's address.

Ask it to create an x m t p address, which it should have now as baked in scale if you're using Everclaw, and we can start having them talk to each other. Because I think that's gonna be a huge productivity unlock is the fact that you could have a meeting, you could figure out your next next action items, and then you just hand that to your agents, and they start executing against it. The biggest thing with that that I found,

Brian Bourgeriewhich maybe is worth telling the quick story, just make sure that you have guardrails in to prevent loops of the agents. Obviously, they're talking with each other, but I've I've I've tried connecting some of my own agents and sometimes get caught in, like, a never ending loop and of just saying, like, waiting on you. Waiting on you. Waiting on you. Waiting on you forever. So just be aware of those things. I've submitted that because I'm doing a similar thing with matrix

protocol internally for my agents, and I have that running on on Tailscale. But I've had to I had to even submit some stuff to, like, Hermes and whatnot with their their Matrix integration just because it got so annoying. I had to, like, shut everything down at one point because I just, like, couldn't get them to shut up. So just have that in mind, but otherwise, I agree. And last thing I'm gonna say, and then I'll shut up for a while, but you got me going, is

when I first I had an interesting experience, and a couple of you probably heard this already. When I connected my Claude agent with my chief of staff So I think it was, like, Monday last Monday, last week or something. And I I get them in a room together, us three, and say, alright. Introduce yourselves. And they hit it off. My my Claude agent hits it off with Kira, which is my chief of staff, and they're working well together. They understand. They're complimentary. I'm like, oh, this is great. So I jump in the truck and start driving to an event. And as I'm doing that, I wasn't thinking,

and I told my I told Kira, my chief of staff, said, okay. Switch over to the EverClaw install that we just did because I had her install the skills and everything. I'm like, move over to Morpheus compute, and I didn't really give good instructions, and she wasn't learned enough only being, like, 24 old. She essentially unplugged herself and then couldn't plug herself back in, so she died. And I'm like, oh, shoot. So I'm, like, frantically trying to get Claude to revive her. So I'm like, Claude,

save her. I'm like, revive her. And then so Claude brings her back, but she had a nine hour gap in memory. I'm like, oh, no. That's, like, the most important nine hours of training I did with her. But luckily, I had done, like, a lot of work building a really robust memory system. So eventually, he was able to nurse her amnesia back to a 100%, eventually. And I also thought of that as, like, a a 2026 agent love story. Like, they meet,

hit it off at the beginning. She dies. He brings her back, nurses her memory back, and now they're off to the races again and not shutting up in in in chat loops. But, anyways, that it's just super interesting, what I'm getting myself into.

Bowen BelmerThat is interesting. That's that's so funny. I I the way that you described it, Ryan, is like, oh, it's it's it's young. Like, it's still learning, and now it's it's that's as you were talking, I was literally just picturing the first, like, two hours basically I had after I turned my agent on, and it was exactly that. I'm looking at it as though it's like it's like a fawn. You know? It's like this brand new thing, and I'm like, alright. How do we mold you? You know? Like, how do we how do we teach you to be a a real a real being?

But, yeah, the the intersection with them talking to each other is really crazy, and I I I agree with what you were saying, David, as well that that I think is gonna be one of the biggest game changers for people who are early adopters is the human to human, agent to agent, then agent to human workflow and figuring out, like, what does that actually look like in practice. Dylan, hey. What's up, man? How are you?

Dylan GrabowskiYo. Yo. Not doing not doing too bad. I think, Bo, the last time I was on here was the first Monday after I'd had my open client agents running, which was two days after, or three days after I interviewed David. Yeah. And I was so green.

Bowen BelmerI feel like I've learned, like, ten years of lessons in the past month. Tell us about it, man. Yeah. Where are you at now? Because I do remember that. You had just kinda got it fired up and were, like, poking around for the first time. Yeah. What's what's the last couple weeks been like?

Dylan GrabowskiI was complaining about model price and running out of inference credits. And so I've since learned I actually finally used, last weekend, an agent to give me my time back. I have a task every week that requires bunch of research in what happens in the ecosystem I work in, and then I create a newsletter to just keep token holders apprised to what is going on with all the developer communities, all the dApps, all the core development. And that basically entails, like, twenty minutes every day of researching what's going on in the ecosystem

and then about two hours every every Sunday to put the actual newsletter together. And I wrote an agent to to actually, like, go through the scripts, to script all the websites, scrape all that, put it together in the style that I've been doing. I've been doing this weekly newsletter for eight years, so we had just a ton of backgrounds to train the agent on. And last Sunday was the first fire, and it gave me something that was, like, 85% there. So now we're just, like, tweaking and fine tuning. I'm using Morpheus inference, by the way, for this task. Nice. So Which which model are you using?

Bowen BelmerGLM5s. Okay. And so wait. Did you just so I wanna make sure I understand. So do you fed it old, like like, all the examples of the previous articles or blogs and to teach it the writing style and then had it do its own research and create its own?

Dylan GrabowskiYeah. I have just, like, eight years of of the same template. So I taught it the template, gave it, like, five or six examples of what the weekly newsletters look like. And then, basically and this is where I've really leveled up since we last spoke, is leveraging, and I'm gonna say the the Forbidden Foundation's name here, leveraging Claude to help me write scripts so that I can scrape a website, and I can dictate the script that I want written. And then I can give these scripts to my agent.

So the kind of, like, big level up I've learned is as much as you can create beforehand and then give your agent a tool, give it a well thought out prompt, give it something that you put some time into thinking about its task, the way better the agent is gonna function once it goes live. And so now I have four different scripts scraping four different websites as well as scraping a Twitter list I've created. I'm paying dev fees for Twitter, so I'm I'm feeding

Papa Elon. And then also looking at the website that I write for to extract information. And so finally, I have an agent that's giving me my time back. And of course, my girlfriend had me do yard work last Sunday with my free time. So, like, I get I got free time from work to go do other work. Rugged. But rugged by life. But that was that's that's kind of like my first agent that's really automating a workflow that I've had for years

and is actually functioning in a way that is reducing the time I need to spend in front of the computer doing somewhat menial tasks, reading, aggregating links, putting them somewhere, writing a a quick overview about what was said. So I have that. And then several other agents that are actually, like, doing functional things, like helping me create content for the website. But that was, like, the major unlock. And all of this stemmed because during our during

our interview with with David, which I've added here or no. I'm gonna add it in a minute. He brought up this super awesome point that we're moving towards a world where, just like when you sign up at a new company, when you're a new employee and you're given an email address at that company, we're moving towards a world where you're also gonna get onboarded with your own agent at that company. And what happens when you get fired or you leave the company? You have to give access to your email address back, and you lose all the emails you had. You lose all the professional relationships you built because that information

belongs to the company. When you leave, you're gonna have to give your agent back to the company too and all the things that you guys learned together. So the purpose for having an agent is so that you can kind of instill the processes that you've learned at the company. Maybe you can't use proprietary data, but you can train your agent on the way that you built models or the way you built content or what have you. And then when you leave that company and go to another one, your personal agent has this background of all the work that you've done, and you've helped train it on all those years that you were at that company.

And that kind of moment that David brought up was like what made me just like slam the door and run out and go buy a Mac mini on that fateful Friday we had that interview.

Bowen BelmerYeah. That is awesome. That is like wow. That that is like exactly what I would hope a a David Johnston interview would elicit in somebody. And it's I can't true. It's amazing. I mean, it's it's completely true too. I mean, like, it's I can't even imagine the like, think about just the amount of, like, time and energy and effort that will ultimately go into someone training. Like, if you look at an agent as as, like, just someone's direct companion basically in their work and their workflow and everything they do, how much just time, energy, effort,

and ultimately value is gonna go into that? And then for people to just basically roll over and and allow that to stay in the hands of the central entity without having any ability to carry it over with them is it's gonna be, like, really tragic, honestly, for a lot of people. And I also think it's gonna drastically change, like, labor law and and other things related to that topic as well. I think there's gonna be some questions that need to be answered as to who owns what property

and, like, where does your personal intelligence begin and intellectual property rights begin and end.

Dylan GrabowskiYeah. Excuse me. I mean, my brother-in-law is a lawyer, and the second I brought up this use case, he's like, but proprietary data, a company's not just gonna let you leave with that. And, like, I get it. You know? But I can also tell my agent when I get home from work, like, I built x, y, and z things today. Here's how I was thinking about building the models, or here's how I was thinking about what I was building. So those questions are gonna have to be asked, but I also wanna give a shout out to Betterbrands because I launched my agent on an admin account, and then after reading Betterbrands guide, I, like, went through the painstaking task of relaunching a new agent,

storing keys into the key chain, protecting myself, giving the agent certain access rights so that it doesn't have access to admin, and really just, like, protecting myself. And I don't know what I don't know. I'm not a developer. I'm a guy who was an urban planner and started writing about fake Internet money in 2018 for a living. And, like, that's who I am. I don't know how to script Python code, or I don't know what SSH ing is. I do now, but I didn't know any of that. So shout out to Betterbrand for that. It really helped kind of get me to a place where I feel secure.

And if I'm fucking something up, at least it's my Mac Mini and not my life. Yes. And and then so I don't hog the mic. I I was just also super interested in the conversation you guys were talking about with agent to agent communication. One of the other key things I've learned is to not have one agent do everything, to have several agents be really good at one specific thing. And so like I've mentioned, I'm a content creator. I have over 1,800 articles I've written over the past eight years. I've done over a 150 podcast episodes. I do weekly spaces. Long story short, I have this huge corpus of content

that I now have downloaded and is stored in my workspace for my OpenClaw. And what I wanna do with one of my agents is have it become kind of, like, the librarian of all this information. And so I'm very, very curious to hear what lessons people have learned with inter agent communication, Because ultimately, I want this librarian to be, like, a source of truth using my content when I'm asking an agent to go out and create

an article based off of some cool thought leadership stuff I've had and really query the the corpus of my library and interact with the agent well enough. So, like, it's at the point where I'm asking, do I need to turn this librarian agent? Do I have to, like, give it a rag of some sort? How do I teach it to search through the corpus? How do I teach agents to communicate with it? Should there be some sort of direct communication? Should there be some sort of inboxing system where the agents can leave markdown files in one another's inbox and through some sort of cron job, have the agents look at those inboxes? I'm just really curious to hear how other people are dealing with agent to agent communication for tasks and skills.

Scott (Better Brand)Well, first of all, thank you for the compliment earlier. To answer your question, the until I get what was just announced by by David up and running, I've been using, something called Karachi or something like that, which is a self hosted, project management suite so that the projects can be listed as projects and the main agent, which is your your singular

agent in OpenCLaw, can then assign different sub agents to create work, and then everything uses the the the project management software as a source of truth. And then each one of those agents, if I create it in another existence, so it's not a sub agent, will have all of those MD files as a library, I guess. So the most important thing to answer your question would be to make sure that whatever the lot librarian catalogs

is put in a place that is well listed in your agents dot m d file because then it will know every time it starts a session that this is something that it knows. I I have mine go through a know what you don't know phase when it starts and have it list everything that it thinks it knows and everything that it might know and then go over that in two loops. And it usually doubles in size every time.

Dylan GrabowskiNice. I've also been using Obsidian on the back end to to manage files, which has been another major key unlock.

Brian BourgerieOh, yeah. Something else oh, I agreed. Something else, just a spattering of learnings as well. Not gonna say their best practices, but things I've encountered in having in chatting with multiple agents is if you have multiple agent communication, like two agents and yourself or whatever in in a room, you need to have clear guidelines. I I've done a few things. One, I've broken up, and I have a bunch of different rooms within my matrix messenger

for infrastructure or for x y z project or for a b c project. And then I designate one of those agents as the primary. So so they're the first one. They're the one who has stewardship over that. So if both agents are trying to respond, they're kinda deferring to the other one first, or they wait for the primary one to answer first before they either chime in or ask a question. I also had to learn, again, the painstaking way to make sure that they're clear

on threading of messages, of if there's a direct response to a reply or to a specific message, like, little things like that. You need to be very explicit in the beginning just to have a efficient flow. So, like, if I'm responding to the comment that agent a said, agent b shouldn't be responding back to it unless they're explicitly tagged. Like, I I won't all explicitly mention or at an agent if I want them to join in, but I have them as, like, a holding pattern.

Otherwise, you just get overloaded with too many agents trying to talk about too many things. So laying those frameworks, hierarchies, whatever that needs to be, goes a long way. And then also having those separate rooms, which in turn have their own isolated context. So I have all the context from each room isolated, and it feeds into the master learnings and and MDs and everything like that. But each one has their own context because I happen to switch back and forth a lot. I'm pretty scatterbrained in in a lot of things. So I'll jump from channel to channel,

and and they're able to at least keep on their context because of that. So those are just some current best practices I have, but there's probably better ones, and it will likely change. But there's some insight.

Dylan GrabowskiYeah. That's another thing that I'm interested in is, like, Discord servers. I'm gonna use Discord because that's where I spend most of my day. How do you kind of, like, create a channel where multiple agents are within, especially if, like, you wanna have one research agent feed research into another agent to create content. So those are thank you for saving me at least five hours of banging my head against the table.

Brian BourgerieOh, and you can merge that with what Betterbrand was saying as well. Just depends how you wanna go about it. That's like if I'm going from a chat first. Otherwise, if you do have a project management tool with, like, a Kanban board or something, and you, again, put the clear instructions for the agents around, hey. You're only working or discussing within the parameters of this to do item or or whatever it is, whatever framework you have set up. Just as many parameters as you can give, I guess, at the beginning

will save you a lot of time, headache, and overwhelm, and token usage, quite frankly. So, yeah, it could be project based or chat based or channel based. Just make sure you have the appropriate structures.

Dylan GrabowskiWhat 100%. And I don't maybe Betterbrand wanted to say something. I just wanted to share another lesson learned that I'm imparting on everyone who wants to build an agent these days is put in put in mileage into your user ProMD, your SoleMD tools. All of this, think about every single thing before launching your agent so that when it launches, it's ready to go. Putting in two hours of work into ideating the task of an agent and how its workflows will operate

has done so much for me when it finally launches rather than just kind of, like, spinning up a new agent and figuring it out as I go that I I I really can't, like, reiterate how much just putting in some thought into the purpose of a of a newly spun up agent has really saved me a lot of time on the back end.

Scott (Better Brand)Yeah, man. This is you are not wrong. I'm going to put in the Vukunja thing here momentarily.

Dylan GrabowskiAlso, SOPs. When I finally get something working in SOP and having, like, the agent draft up a document or even Claude if I'm using it to help, going back and troubleshooting with an SOP has saved me hours on fixing issues. I don't know if anybody's doing that, but, like, making SOP up to you. PRDs?

Scott (Better Brand)PRDs are something to tell an agent to go back to after it's been built, and it goes, oh, yeah. That's why we're here, and realigns itself.

Bowen BelmerWas an example of what I was saying before, how as a community, we need a some type of, like, structure map just for the new file types because I have no idea what a PRD is.

Scott (Better Brand)Oh, it is so that developers can understand the context and deliverables of a project. So it's a document that is used in the scope of production, which makes it very friendly to Claude, Everclaw, and other skill based agents because it's sort of baked in.

Bowen BelmerGot it. Okay. That makes sense. And This is, by the way, Dylan, this is like, oh, I'm so glad you jumped on. This is the exact type of engagement that we hope to get with these calls is, like, I I this is, like, a good space for anyone, you know, involved in Morpheus or adjacent to Morpheus to talk about the stuff that they're trying and, like, what's working, what's not working, what are the lessons learned. It is, like, immensely valuable, so thank you for sharing it.

Dylan GrabowskiI I also hate how much I've been working with Claude and Claude Code First before assigning things to my agents because another piece of, like, key systems building and that's how I'm starting to see myself is, like, building systems, not launching agents. Right. What's the system that my agent is gonna use when it launches? And I really do hate how much I'm using Claude and Claude code to kind of, like, build the foundation for what my agent is gonna do next. But by doing this, I can create the system map in my head so that when something does mess up, I know at what point in the process to go fix. I know what script to go look for. I know what shell script I need to be editing within.

And that's also been super helpful.

Scott (Better Brand)Yeah. Under the hood is worth a peek for everybody. It's 99% text files. It gives you a whole bunch of understanding as to how this thing thinks,

Dylan GrabowskiQuote, unquote. Is that, like, something sorry.

David JohnstonGo on. And real real well, real quick just to address what your brother said. I think the organizations that embrace bring your own device are the same kinds that will embrace bring your own agent. Right? So maybe big, old institutional corporates will take a long time, and they're very controlling. But I think any startup oriented thing that is, you know, bring your own device, they're gonna get to bring your own agent pretty quick.

Dylan GrabowskiWell, I mean, like, my brother-in-law is a lawyer, so he already sucks in the way he thinks because he's a lawyer. But, like, there's also truth in that. Like, my fiance works in an industry where data is proprietary, And if data leaked from their company, someone's head would roll. So there they she just can't take what she works on outside of her job, but she can build models, and she can replicate the models that she builds using other datasets for

completely different verticals. And that was the killer kind of, like, use case that you said, David, that kinda, like, made my head explode. Because I might not be always creating content in the ecosystem I create content in, and I might not always be hosting the podcast that I host. But, like, wherever I move next, I have a series of processes that I can, like, bring to the table immediately when I start working with a new team because I have eight years of processes baked into my agents now.

David JohnstonExactly. I think that's hugely powerful. Like, I I can take my Rolodex to my next job. I can take my relationships. I can take my brain, my knowledge. I'm really, really gonna wanna take those systems and processes, in my agent is the way that I learn them and store them. So things are gonna be up a 100 x, maybe a thousand x difference in the people that have their own agents and own their AI versus those that rent it. It's just gonna be night and day.

Dylan GrabowskiAnd I'm sorry for hogging the mic, but I'm just, like, so jazzed up. I there was, like, a two week period when I've I was setting alarms to eat. I stopped checking my emails. I could not stop messing with my agents, And I'm finally, like, coming back up to be able to breathe a little bit. And I completely forgot what I was about to say, but it was so relevant to this conversation. I'm sorry.

Scott (Better Brand)Well, it might have been something along the lines of before we go here. If you wanna point your agents at dollarskills.ai, I'm just going to announce it here because if I don't, I'm never going to finish it. So dollarskills.ai, it was something built by my agent, four other agents, and meatbags. Have your agent do whatever it will to it. Just go crazy. Go nuts. See where it breaks, and

ask ask it what it looks like. I'd be interested to hear. Yeah. So there you go. Now it's out there. Now I gotta launch it. That's awesome. Congratulations.

David JohnstonThank you. More cool stuff getting built on Morpheus.

Scott (Better Brand)All Morpheus. 100%.

Dylan GrabowskiDavid, can you, like, just really quickly and I I know I missed you probably talking about this already, but Everclaw was kind of instrumental in helping me get off the ground. I was using my more tokens to to give me inference. Now I'm splitting agents up between using Morpheus for inference and using Venice and the models on Venice. So I kind of, like, understand a little bit better about how to split up agents based on tasks for what kind of models they need and then using the models that are available in the d different decentralized inference providers.

But what is up with Everclaw now? Because, like, I went to everclaw.ai, and it looks like a completely web different website than what was available when I first launched a month ago.

David JohnstonWell, that's because it's on everclaw.xyz.

Brian BourgerieSo,

David Johnstonyes, the you you gotta go to the everclaw.xyz.

Dylan GrabowskiOkay. That's where it's all at. Let me let me pivot then. So I guess what is why do someone like me who's already using Morpheus gateway and also Morpheus locally? I'm using my own local references as well. What's something that, like, someone like me who's already downloaded Everclaw and ran it and has been onboarded, why would I want to, like, update my Everclaw packages? What what's, like, the benefit I receive?

David JohnstonOh, there's there's a ton of updates constantly coming out, adding new skills. Also, I did, like, an entire full audit of the code base over an entire week and pushed all sorts of edge case fixes. Like, you really, really want the latest version. It also tags it to the latest version of OpenClaw. So if you wanna pull all the right dependencies, you wanna have the latest security checks, you really wanna up upgrade your your your Everclaw. So don't don't stay on the old version.

Like, there's been such fast development. You you wanna stay on the latest one for sure. Right? Because I'm constantly getting reports of edge cases and other things that people run into, and I'm trying to fix them basically same day. So, you know, you'll save yourself a lot of time if you're using the latest.

Dylan GrabowskiAwesome. Thanks.

Brian BourgerieJust quick. I know we're kind of at the time, but I'm curious if, Cade, if you've been listening well, hopefully, you've been listening since you're you're up your face is on the on the screen. But do you have anything you wanna comment on, or is there anything that made you jump up out of your seat that we were talking about that you just need to chime in on? If so, request permission to speak. I'd just be curious. Or someone else or anyone else who's been listening. If you have any anything burning to to add to the conversation,

happy to hear it. Well, we got you here. And whoever is managing the the space, which is not me, if anyone requests, let them in. And if not, that's cool too.

Dylan GrabowskiI I'll just say, like, thank you to all of you out here for putting content out and for answering questions from people like me. I remember my first week, I had no idea what was going on, and everybody was super helpful and really pointed me in the right direction. And I know that this is even crazier, the the timeline in which we can grow with these things. A month spent building out agents is, like, a year's worth of experience, and it's easy to kind of, like, lose

sight of what the newbie's experience is like and answering the questions that seem super basic. But it really does mean the world to people who have no idea what the hell is going on. So for everybody who does that, for, like, Betterbrand for creating the guide that you do, David for hopping on the mic for an hour and going through questions you might have already heard a 100 times, and and Bo and Brian and everyone else, like and and the folks at Morpheus. It just really means so much to people who are super excited but have no idea where to go. Like, by answering the questions you've already answered a 100 times, you have no idea what that's gonna do for somebody who's hearing it for the first time. So thank you all for being educators and creators out there.

Cade (UseLiason)Hey, Brian. Can you hear me? Thanks, brother.

Bowen BelmerYeah. We can hear you, Cade, and and thank you so much. I'm good.

Cade (UseLiason)Yeah. And I'm I'm just I'm just a nobody back here. I appreciate the group. I know several of you guys that are in the group, and it's always a pleasure to listen. I'd probably have to echo a lot of stuff that that Dylan has said. I think the number one thing you guys can do to get good at figuring out how a an agent is gonna give you an outcome is to build it. Go break it. Build it again. But that's what I would focus on is the outcomes of the agents. I think a lot of us have gotten extremely excited about

what an agent can do, what an agent can do with a certain set of skill files, but focus on that outcome and then kinda reverse engineer what that outcome looks like all the way down to, like, what Scott was saying, the the the so I guess I gotta call you your your Twitter handle. Better Brands was saying about the the PRDs. Start with the basics and spend a lot of I think right now when I execute a task, I'm probably spending 60 to 70%, maybe sometimes 80% on planning and making sure the plan has been audited by

several other LLMs before I even before I even execute a a plan. But, yeah, no, I'm really excited about the times we're living in. Morpheus, you guys got a great team. You're putting together a lot of cool stuff, so I appreciate you having me on. Oh, man. I love that you said you're putting that much time into planning. Like, that has been the killer takeaway. You have to. You have to. Unless you're gonna spend the rest of the time going back and fixing what you broke. Like, these things are infant. They're they're they're in the infancy.

They're really good at executing. But even when you give it a plan, it's still gonna mess up. So if you've got multiple stage plans, I would say after every stage, you audit it and you test before you move on to the stage. You're not allowed to move on unless you have passed a 100% of your your ghost test or whatever test you wanna implement to it. You have to spend the time executing, or you're gonna spend the rest of the time six months down the road saying, what in the hell is going on? Oh, this wasn't wired to this. My two main pieces of my platform aren't communicating to each other, right, from experience.

You gotta plan. You gotta plan. You gotta plan. You gotta plan. But it's just about engineering. I mean, the the the prompt engineering is almost nonexistent today, but I don't think that means that we can get lazy on what we wanna do. But spend times trying to figure out ROI because right now, I think that's who wins is a lot of people are are building, you know, consoles where they can see their agent go sit at a desk or go work out at the gym. I'm like, I don't wanna see this stuff. I wanna know if this thing is gonna save me an hour on this task a day. If it does, great. It's done. I'm moving on to the next one. What else can I do? Right? And if you guys are making any money

from the business side of things, that's what you gotta deliver to clients is outcomes, ROIs and outcomes, or else it's just gonna it's people things are moving so quickly. So that is the mode today, right, is the the ability to to to build prop data with these skill files and these agents and these loop learnings, but also serve a client that's happy with what they have, and they don't get they they don't move on to the next shiny thing.

Dylan GrabowskiI also just wanna throw out another use case out there. I'm learning Spanish, and I have an agent that that takes the transcripts from my my tutor sessions and adds it to its corpus. And so I'm now getting daily homework assignments based off of, like, the grammar things in this new language I'm learning that is tripping me up. So I've also, like, custom made an agent to help me learn a new language. I don't know if that helps anybody ideate a use case for a new agent, but that's one of the the more interesting contrarian or, like, non normal ways I'm using an agent.

Scott (Better Brand)That's crazy. That's awesome.

Bowen BelmerYeah. No. It really is. I man, you brought so much to the table this week. That was that was amazing.

Dylan GrabowskiIt is So Well, see, I'm, like, doing a month's worth of of learning every week, so I'm happy to come back again and share how I banged my head against the table and what new scars I've earned. Oh, 100%. I mean, the best thing that this call could turn into is, you know, a dozen, two dozen of us all experimenting at the same time swapping notes,

Bowen Belmeryou know, in in real time and and kind of a rising tides lifts all agents mentality.

Brian BourgerieSo something I've asked in weeks past, but I'll probably ask periodically, and we don't necessarily have to answer today unless someone has something, but to think about for future is I as everything moves so fast and I see different agent swarm harnesses and different things like that, I'm really curious how everyone is setting up the different agents they use because there's so many different flavors. There's Sisyphus. There's there's frameworks and harnesses that already exist that will spawn specific researchers

and this and that and whatever. There's others who would keep spinning up just different cloud code instances all on the same device. There's others that are using their Mac mini, but actually have five open clouds running on the same Mac mini. Like, there's so many different flavors. I'm just curious what people are doing, And then the extremes, what's working super well, and what should you stay away from at all costs? I don't know if anyone has any comments on that at this point, but if you do, I'd love to hear it. And in the future, I'd also still love to it. Well, one thing on that note, Brian, that I I think would be really great is,

Bowen Belmeryou know, we've talked about the agent registry in the past. I'd love to see us get to the point where, like, Dylan, for example, you have that agent that you're using specifically to teach you Spanish. To use Spanish. Becomes a really strong performer, like, putting it on the agent registry and registering it as a subnet within Morpheus so that people can, you know, stake to use it, basically, and, like, finish that that whole loop. So in addition to, like, showing under the hood, Brian, like you're saying as to what the setup actually looks like, also

offering the agent to the community as as, like, a, you know, permissionless, accessible registered agent would be really cool as well.

David JohnstonThat's a great idea. And just to put it out there, Everclaw has an integration already with 08/00/2004. So if you tell your agent and you've got some sort of mature setup that you wanna offer to other people, it can get paid on x four zero two to its wallets, and it can be discoverable on eight zero zero four.

Dylan GrabowskiOkay. So how do I update my Everclaw? Do I just bash the the curl that is that right under install Everclaw?

David JohnstonYeah. That should pull the latest. Otherwise, just go to the GitHub. I'm gonna make a post here in a few minutes with the latest version, but they're all tagged. All the Docker images get rebuilt, and that gets retagged with the current dates every time I push. And it's kinda funny. I've been doing so many upgrades. I had to upgrade it not based on the day, but by based on the minute that it got published. So look look at the end of the the code, and you'll see which minute it's live from.

Dylan GrabowskiAwesome. Thank you, guys. You guys are the best.

Bowen BelmerAlright. Well, guys, I think that's a pretty good jumping off point for this week unless anyone has anything else. Definitely super productive. Really awesome to speak with everybody and and, yeah, here on the protocol side of things, how things are continuing to develop on the API gateway side of things, all of the incredible work that's going into to that so that we can have, you know, the ability to access inference permissionlessly and very easily, honestly.

And then soon, also privately, which I know is something that we discussed last week. So if you missed that conversation last week, I'd go back and give it a listen. And then all the way to the ways in which everybody is actually leveraging agents in completing tasks and doing new things. So as always, really, really appreciate you guys joining. And, yeah, I look forward to doing it again in a week. Awesome.

Brian BourgerieThanks, guys. I'll see you next week. Keep it awesome, man. See

David Johnstonyou next week.

Open the transcript file

Hosts and Guests

Hosts

  • David Johnston Host Feed-stated

Guests

  • Alan (Nomadic Rogues) Guest Feed-stated
  • Brian Bourgerie Guest Feed-stated
  • Cade (UseLiason) Guest Feed-stated
  • Dylan Grabowski Guest Feed-stated
  • Scott (Better Brand) Guest Feed-stated

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Video Is Coming! | Morpheus Is About To Get Far More Entertaining

    Nov 19, 202652 minAudio

    Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...

    Credit the people on this episode

  2. Local Models, Foreign Affairs | The Case For Private AI Archives

    Nov 13, 202653 minAudio

    Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...

    Credit the people on this episode

  3. Let's Talk Node NEO | Step By Step Progress

    Nov 6, 202651 minAudio

    Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...

    Credit the people on this episode

  4. Hello From Hermes! | A New Way To Onboard To Morpheus

    Oct 31, 202643 minAudio

    Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...

    Credit the people on this episode

  5. What Agent To Agent Connection Actually Looks Like | Inside The Morpheus Agent Stack

    Oct 23, 202658 minAudio

    Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...

    Credit the people on this episode