AI Research OS | Outages, Open Models & Real-World Workflows
44 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 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 #58! Recorded: 15 January 2026 AI SUMMARY: This week's builders update highlights Morpheus developments including updated API docs and extended free API access. It explains that scale will come from agents and workflows not simple chat and demonstrates building AI Research OS by switching from centralised providers to Morpheus' marketplace. This mixed specialised models for better depth efficiency and latency outperforming centralised tools. The update also covers inference economics privacy needs and security concerns. It advises new builders to start with n8n and architect first then implement. Timeline: (00:00:00) Intro (00:00:39) Morpheus updates: docs, app and billing timeline (00:05:40) Introducing AI Research OS and the deep research gap (00:08:43) From centralised to open: Venice and Morpheus switch (00:10:29) Choosing models for cost, latency and quality (00:15:45) VC subsidies, data trade-offs and excess capacity (00:18:03) Privacy as insurance and looming "oh no" moment (00:22:12) Phishing, deepfakes and on-chain verification (00:26:11) What tools the team uses for privacy-first building (00:29:05) AI Research OS results vs GPT, Gemini, Perplexity (00:33:26) Idea people, infra realities and a live build idea (00:36:37) Regional regulations: bans and access workarounds (00:39:32) DAI adoption across regions and open models trend (00:41:56) Closing: next builders, workbench and thought leadership 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 JohnstonGood. Well, I'm glad the Internet outage apocalypse twenty twenty six has not knocked you out. I guess Verizon or all these different infrastructure has been having trouble yesterday. So if anybody's been experiencing slow Internet, it's not just you.
Kyle (Bow Tied Blue Fin)It's it's all over the place. But looks like it's come back. So, Kyle, do you wanna kick us off, man? Sure. Yeah. The that Internet impacted me yesterday, and and I like, everyone thought it was just me. So happy that we're we're back, and it's funny how quickly the world can shut down without Internet, it seems. But yeah. No. Excited to go through the builders update today. Today is going to be similarly following the new format that we've been doing for this, but a little bit different that I'm actually going to be giving the update
as a builder today. So before we get into that, we like to start with just some high level Morpheus updates on the last week. I'll say that the Morpheus API documentation and app.more.org has been updated again. So we know, you we're just in the process of cleaning up all those docs and making sure that, you know, you really get the best experience when you go there with up to date, easy to use information and guides. We'll probably probably be sending out some tweets today or tomorrow on
some of the updates there. Next next important thing is, again, on API billing that we mentioned every week. We are really progressing now, and and we're really rolling with some of the infrastructure on our side there. We still commit that everyone will have free access whether that's through, you know, billing not being turned on yet or by just manually assigning credits until the end of the month at least. So really excited, and we think that that will gonna be a big demand driver for us from both the consumer and provider perspective on on kind of bringing the scale of the inference marketplace to the next level.
Now if we think about how does that happen, right, how do we get scale? We do that by people building on top of the system. And I I don't think it's people chatting that's gonna drive the scale. I think it's agents and and workflows and software that's going to kind of drive up those inference needs. And we heard that last week with was last I believe last week with Tales and con conquest or maybe the week before. We we continuously hear different people building and and running their software through
an easy to use API like Morpheus, and and I'm gonna talk more about that in a second on something I've been building. Before we get there, Brian, Beau, Thomas, David, anything you guys wanna talk about just kinda last week in Morpheus on on the highest level without getting into kinda the the thick weeds of it?
Bowen BelmerYeah. I have a quick just really quick thing on my side, Kyle. So the first is that we are at the very final touches of, Brainpower season two episode one. You'll be seeing it very shortly here. You know, thank you to everybody as always who who does listen and engage to the content. Please, give us, me, David, some feedback if you like the new format or if there's anything else you'd like to change about it. And then the only other thing I wanted to add to what you just said, Kyle, with regard to this call is for people
who are maybe more familiar with, like, the DeFi space than they would be the DAI space, what this call basically is and what we see it turning into is if in 2016, '17, and '18, you were attending Ethereum foundation calls and you were seeing projects like Uniswap and, Compound and, like, these early decentralized apps who are building on top of Ethereum coming on each week to talk to you about the things that they're building in practical terms and how they're doing it and how you can leverage
the technology to build your thing or interact with their thing. That's that's really like what what this is and is meant to be going forward. So you know, ultimately, I believe that the the killer app or the killer apps, or I guess agents in this case that get built on top of Morpheus will, you know, you you will maybe very likely hear hear about it first on this call, which is which is very cool or at least from the team directly.
Kyle (Bow Tied Blue Fin)Yep. Couldn't couldn't agree more about it. I I think that's exactly kinda what we're looking to promote and nailed it. And we are all very excited for brainpower season two as well. So can't can't wait for that to come out and start listening.
Scott (Better Brand)Yeah. I love the stories kinda putting together of the, you know, people enter this space. And, you know, you you kinda start off with this kind of centralized centralized options, and and then you kinda expand from there. So it's really great to see, you know, you know, Kyle, Boeing as you guys are each running through, you know, from different angles. You're showing the stories of individual builders of what they're going through, kind of the problems that they faced, the pain points that they faced, and and why they're turning to open source models, why they're turning to,
you know, what what Morpheus can help with them. So, you know, kind of in in that vein, Kyle, you wanna run through kinda kick off what you've been working on. What's what what what have you been working on? Why did you get started on it? And yeah, what what can you tell us about it?
Kyle (Bow Tied Blue Fin)Yeah. For sure. Can you guys still hear me? I don't see my mic popping up. You're you're great. Yeah. Good. Cool. So I've actually been working on a project called AI Research OS, and the the problem that I was dealing with is that kind of as as a user, I wanted to do deep research, and and I wanted AI to go and search the web with with all these sources and kind of come back for
it was a competitive analysis on different AI labs versus each other for for model performance. And when I ran it through GPT, it was decent. Right? Like, GPT does a great job of grabbing a couple sources, conducts the web search. If you click the deep research, it takes a little bit longer, but it wasn't really to the level that I was hoping for. I did the same thing on Perplexity, the same thing on Gemini, and, like, not gonna lie. Right? It's good for for most basic cases, but anything that you really want a comprehensive analysis of, it just seemed a little lackluster.
So I so I thought that that in the age of today, like, we we can do better. And so what I did was I I went and I got it started in in n eight n like anyone would for for a simple kind of architecture to see how how something works, and I just plugged in GPT. Right? I grabbed a a GPT node, put in my my API key. I connected it to a couple different nodes within there to conduct web search, you know, Google search and then web scraping and and provide a simple answer right on the most basic level.
And and it worked. Right? You're you're able to generate, you know, requests, search the web, and and get an answer back. That's one of the most simple kind of any of them workflows that you can do. And after that, when I started kind of getting more into the needs, right, like, are what is deep research for? What are people going to use this for? A lot of people are going to use it for market research, for user research, for identifying the proper features or competitive advantages or disadvantages or pricing. Right? All of these things that require
deep context and deep understanding. And to do that, you really need kind of recursive flows where you have a search. Right? You learn something from it, and then based on the answers that you get, you wanna go a level deeper. So now it's like, you know, what color is the sky? Alright. Well, the sky is is blue, but that's only during the day. So alright. So what color is the sky during the day? What color is the sky at night? Right? And it's kind of like a second search. And then the third search is, alright. Well, is the sky different in different parts of the world? And is and you kind of go deeper and deeper and deeper, and that gives you
more granularity in your request and a more contextual answer when you're trying to draw a conclusion from it. Maybe that wasn't the best example, but we'll we'll move on. So once I implemented all those, I thought, does this really need to go through a centralized provider? That's good for testing and and first case, but I like this to be uncensored. I don't want someone to be able to block my access. I don't want someone to be able to steer where the results are are going or have any bias. So or I really want this to be fully open and and fully transparent and, like, freedom of access to information on the Internet. It's not it it's something that should be
trivial, but it's not. So I went in and and I took my OpenAI node and I at that point, switched it to Venice. So, you know, we we know that Venice is a one reputable provider in the industry. I wanted to see what happens if I switch there. So I went in, changed from the OpenAI base URL to the Venice base URL, put in my Venice API key, grab the models list, and and kinda just went with it. And
within, you know, three to five minutes, I was able to make that transition there, and it all worked properly. From there, I said, now I need to switch it over to Morpheus. Right? Kind of walking through the stages of how I think people will go through the process from fully centralized to fully decentralized. Ninety seconds or or less. Right? You you know the base URL. It's in all of our docs. You're able to grab an API key in a couple minutes on app.more.org. And the tools like any then and other tools are able to automatically
fetch the available models from our models endpoint, expose them to you, and you just select them. And some of the real benefits that you get from using decentralization here is you get the super wide variety of models that are available within the marketplace. Right? I have some Quinn models. I have GLM models. I have Lama models. I have all of these different model types that are good for different things because there's not a one size fits all. Right? We know GPT is is good for some specific task. Gemini is better at others. Claude is good for others,
like coding, for example. So I chose, I'm able to pick and choose my models. I choose a super small quen model for initial categorization of data, and I use a massive GLM model for kind of reinterpreting learnings and and developing conclusions from them. And I use a middle tier model to generate search queries based on what's been learned so far. And when you're able to plug all these things together, you can create more you know, get closer to an optimal workflow and and really optimize both cost and latency and time and performance and everything as you go as compared to being stuck to a couple options that,
you know, are inherently centralized in something and that you're giving your data back anyway. So I know a lot of words there, but really just my experience in how, honestly, how easy it is in these days with tools like any of them and other tools to just flip over from centralized providers to using something like Morpheus. It's just a couple minutes, and and you're in and running on something that that you can trust and something that, you know, you can be proud of that that's running your platform that you know you're protecting your your users who who you're making a commitment to, and you're protecting your own data as well.
Scott (Better Brand)Todd, you'd said something earlier on the you'd mentioned before several times that, you know, the, you know, the availability of compute, you know, while it was free now with with a lot of models and and there's a lot of it there's a lot of incentives from venture funds, you know, to be able to get cheap access to AI. What are you seeing now? And you you described that that access to AI, access to compute was was gonna be a bit like an oasis in the desert where people are just kinda starved and fighting over themselves for for access to toe you know, tokens.
How are you seeing that involve, evolve in the market right now?
Kyle (Bow Tied Blue Fin)I think it's an interesting time in the market. Right? Because a lot of people are being tricked that here here's some free tokens in a lot of cases. I mean, obviously, not for us. We're just losing money, right, by by giving away free tokens. But if you go to, like, DeepSeek, for example, it's like, oh, get, you know, 1¢ per million tokens if you use us. All you have to do is KYC, give up all your your personal security, let us train with your data. And I think that there's
the market is turning a little bit to understand that, hey. Maybe this isn't the best method, and maybe we have to pay attention a little bit more to who we're giving our data to for training purposes. And I think that what we're going to see is more of a kinda like a a a distribute I guess it's a binomial distribution of, like, token pricing for for
private data and and token pricing for data use for training. What you mentioned before about VCs funding tokens, I mean, we don't know. Right? It we do see money still going into OpenAI and Sam Altman saying that pro plans are losing money every month, which is only sustainable for so long. So I think there's probably market impacts here on what's offered by VC backed funds who, you know, do have major interest in storing your data. And then
on the other side, we also have the aspect of open marketplaces where many many, you know, companies are coming out there who are building massive data centers, and and they're going to have excess capacity. And and what we see is the opportunity for utilization of some of that excess capacity in the Morpheus inference marketplace. And for a lot of these people, it's going to be, how do I capture some incremental revenue on lost capacity, right, you know, that that I wouldn't have gotten otherwise. And and we believe that
that will result in much lower pricing and and kind of give a very compelling pricing story to Morpheus users in addition to the privacy decentralized, you know, more key value propositions that that we see right now. Did that answer the question or was that kind of did I just walk you around the the block a little bit there?
Scott (Better Brand)Both and. It's you know, I appreciate
Bowen Belmerit. Sayeed, what were you gonna say, Bowen? Oh, I was just gonna say what I one, I think it definitely answered the question, but what I found so interesting about your explanation is it was proving the concept strictly based on economics without and then right at the end, mentioned, oh, by the way, this is not even accounting for the fact that you have the, permissionless access and and privacy, which I I would venture to say that that bonus or, like, benefit of I guess, features maybe the right word. That feature of what Morpheus is is not really being priced into the market yet in my opinion because we haven't yet gotten to the point where you're seeing,
large scale government restrictions on AI businesses and, like, where you can go to access inference. But I believe it's coming. I mean, you now see The UK essentially threatening a ban of x, you know, within their borders, which would probably include grok, I would think. And and that's a slippery slope, right, just like in general. And so your answer like provides why it's it's such a good opportunity for builders
strictly economically without that market force. And when you add in that market force, it's like orders of magnitude probably in terms of what it means for the the economics and the tokenomics.
Kyle (Bow Tied Blue Fin)100%. I I think that when you it's kind of like if anyone remembers the early days of of AWS where why did AWS win in the beginning? Because it was the cheapest and it was the best. You know, it was they kind of said, we're going to just go all in on our users, and we're gonna make the best product that we can. And I think for us in this situation, the inference marketplace has the opportunity to win
on the pricing perspective standalone and on the privacy perspective standalone. And when you kind of have different users with different needs, you're able to kind of capture a wider subset and have a wider kind of addressable market of people who are interested in the product. I think that the privacy side is something that will become more critical over time. Right? We just need a couple more not that we need. Like, I I don't want this to happen, but
if OpenAI accidentally exposed user data in more cases or like we saw with the the whole Facebook llama fiasco where where prompts were being posted on on Facebook profiles, and and we see this all over the place. Privacy right now is insurance that a lot of people don't wanna pay for until there's an accident. And then it's like, oh, shit. I yeah. I need this. I I can't really live without this. And I think that we're not at the oh shit moment yet, but it's coming. And I think that Morpheus is in the perfect position
to to to be there when it happens. And and I'll save that I'll just leave that as a key into next week when we hopefully have a little bit more to discuss on some things that are being worked on on the privacy side. But I think we're we're kind of we have the right view of of what the market needs right now, and we're progressing properly towards, you know, directionally.
Thomas BorrelYeah. Just to add one thing to to what you said, and by the way, I completely agree is you've talked about the issue with LAMA and some of the of prompts ending up posted in Facebook profiles. But you also have to think about the tool chain that the centralized vendors are using. And I'm sure you've all seen that in November, Mixpanel had a security incident. And through that security incident, OpenAI got affected because they're using Mixpanel to
monitor the experience of their users and so on. Right? So you have, typically with these centralized solutions, you have a variety of tools embedded within the tool chain that add to risk of the user information ending up someplace in the dark web or or exposed
unnecessarily. Well, not that it would ever be necessary, but you know what I mean.
Kyle (Bow Tied Blue Fin)For sure. For sure. It's it's like your risk like, not the umbrella, but the the degree of risk that you have and and how far it expands with different platforms.
Scott (Better Brand)I mean, I think I think we started off the call, Ray, talking about the you know, you touch base on on kind of the Verizon, outage, was it yesterday that you you went across the entire US rate, and, and and you don't see don't see the risks until it until it goes down. Right? And I I think, you know, Ledger had this you know, Ledger's had these leaks. We've had these massive data leaks for contacts where where, you know, criminals are able to track down information, you know, about where you live and and some rough information about where you shop. But but is I I do think that, you know, Kyle and Thomas, to your point, right, is you get these leaks of,
someone, you know, being able to track exactly what your thoughts were, you know, all the information that's that's gone on with, GBT and and and Meta, you know, all those thoughts be getting leaked out there. You you to to your point, you don't you don't care until you do. And then, generally, when you do, anybody that's been a victim of of of, you know, serious identity theft understands, you know, kind of the horrors that that, you know, that kinda come with that. So it'd be interesting to see interesting and a bit bit chilling
to see that play out over the next, you know, twelve, twenty four, thirty six months. When, not not if, but when, you know, massive amounts of of of hyperpersonal data gets out there and Sure. You know, how it gets used by
Kyle (Bow Tied Blue Fin)bad actors. Absolutely. And think about how much easier it is it it will be to do phishing attacks if you are able to get into someone's brain and how they think and how they act. Yeah. I I fully agree with you though. Oh, totally. I just saw
Bowen Belmerthere's this YouTuber named his channel's I think it's Tommy G. I don't know if you guys have seen this guy. He actually, he's from Milwaukee, so shout out Tommy g. But the guy he was the Kia Boys documentary guy, if you've ever seen that. Yep. Yep. So he just recently put out a video, like, eight days ago that I would highly recommend everyone listening to this watch called Inside a Teenage Prime Empire, the the Billion Dollar Theft Ring. And he interviews,
head of security at Kraken and also Coinbase, but then he actually has interviews pseudo anonymously with and when I when he says kids, like, some of these are literally 16, 17 year olds and then some of them are are adults, with ties to like actual organized crime organizations. But they're basically highlighting crypto crimes and crypto scams. And I think everyone in this call who's been a a crypto user for a while, I'm sure receives Coinbase scam, like spam stuff, you know, constantly.
And it highlights those operations and and like how they function. And it is honestly terrifying. Like, it is it is honestly like one of the first times I've seen media like that and been kinda like, man, you know, double checking my own op sec and just really, like, rethinking through all of those things. And when when you add in data leaks, just from your your Ledger account or your Coinbase account, but now every single conversation you've had privately with a large language model or multiple large language models,
the profile that these people will be able to build around you are just yeah. It's it's very, all encompassing and and in some cases like there are literal children figuring this out, so I can't imagine what you know, a seasoned criminal will be able to come up with. It's it's a little haunting, honestly. What what was the name of what was the name of that that interview again? I'll actually I'll post it. I'll I'll reply to the spaces, and I'll drop the link in there. Cool. Because, yeah, it has the head of security for both Kraken and Coinbase in the video talking about
this topic exactly, basically.
David JohnstonWell, appreciate it. And it's a good reminder because I don't know if you guys saw some of the the latest deepfake videos that hit x yesterday, but it's photorealistic. It's real time. Person's moving their head. The avatar that they're representing as is moving their head and speaking, and it looks perfect. Right? We have crossed the uncanny chasm, and, you know, it is it is, like, perfect now. Right? And so
you're gonna have an explosion because we've been talking about this for a year or two. Like, the video deepfakes are gonna get good. They're gonna get good. It's here. Right? And so you're just gonna have an explosion of scammers going after, especially older folks, but really everybody that's not aware that this technology has reached that level. Right? And so I just for me, it reemphasizes why these tools are so important is we need to figure out how to actively protect people from those type of attacks. Right? And if you're not using on chain tools, it's really hard to know what's real. Right? The the beauty of blockchains is as an append only database,
its history is written in stone. Right? And you can't fake reputation. You can't fake all these things. So, you know, I think it's just gonna accelerate and and create this race where we we need these tools faster and faster. Because on the other side, there's plenty of people that wanna take advantage of of the capabilities.
Scott (Better Brand)David, you brought up a good thing there, tools. Thomas, I think you brought you'd brought up the tool chain. What what what tools are you guys using right now as it pertains both both on the privacy side, security side, and and within within AI? So two two questions. What what are you seeing? What are you liking? And why? And then the second question to it is kind of a follow-up to it is how are you ensuring that these kinda fit within the model of privacy that you're that you're looking for?
Kyle (Bow Tied Blue Fin)I would say for myself, the the start is anything that I am hosting. Right now, I'm trying to run through decentralized providers. So I'm using Akash for a lot of hosting needs along you know, even for any then, for example, that I mentioned earlier, running some early workflow infrastructure, self hosted through Akash decentralized inference just to ensure that there's kind of complete privacy there.
Of course, using, from an AI perspective, using something like Morpheus to the models or or other major decentralized providers in that nature. I do use centralized providers here and there for specific activities for comparison purposes or to try and get a different, you know, perspective on on something for test results, but, generally, all, you know, decentralized AI in that matter. Think
from the other perspective, I guess I can I'll let someone else jump in. I'm not I'm not sure where else there is from the privacy perspective there.
Thomas BorrelI think the, you know, the the other one really, which is sort of the the underpinning of all this is some high degree of paranoia. Right? And every time I look at a tool, every time I look at a functionality of a tool, just wonder if I really need it or not. Because there's that there's always you always have to strike a balance between convenience and privacy.
And so that's something that needs to be top of mind in in my opinion and part of your normal your normal ops act.
David JohnstonI just wanted to say I also see that Eric from Morpheus Asia is on the line. Thanks for jumping on, Eric. I know it's late your time. Thanks for representing the Asian time zone. Hopefully, we can get him promoted and, yeah, hear if there's any update from all the builders over there.
Kyle (Bow Tied Blue Fin)Great. Yeah. Before we jump into Eric, I'll just do one last plug to to close off my talk on AI research OS is, you know, to pulling that altogether, all of that model manipulation and choice of different models and and the types of workflows, it it got the product to a point where it was about 10 x more sources than GPT. And when you run the request against one another, I did
Grok independent evaluations, and I found that research using AI Research OS was more comprehensive, and it recommended AI Research over AI Research OS over GPT GPT Gemini and Perplexity in multiple different applications. So it's just, I think, a big win where you can beat centralized providers with decentralized open source providers. And
I I ask everyone to go give the product a test and and let me know what you think. There's a a free tier available, and you can always DM me for if need any help or or want some credits to do a specific test that you wanna get feedback on.
Scott (Better Brand)I think one other thing, Kyle, you've been mentioned before. You I mean, you you've been in the industry for a little bit. You've kinda done the deep dive, kinda all the tools that you're using right now. Where would you advise a you know, somebody that's kinda just starting off with you know, after being in the industry for a little bit, where would you tell somebody to start?
Kyle (Bow Tied Blue Fin)I would say getting started is really well, two things. First is just just get any then. Pay the $20 for the cloud version to get started just, you know, while you're playing on your laptop. And then more importantly, get out a a piece of paper. Right? And and it's it's all about architecting what you wanna do. Right? It's once you understand who your user is, what you're looking to build for them, what's important for them. Right? What are the key requirements that that I need to do to build this product? Then the building part is not necessarily the hard part. It's the fun part where you start to plug things together in different ways and and and start to self optimize there. The hard part is really getting started. So I would say in in right now, the idea guys, right, and the thinkers are are becoming the
are surprisingly becoming some of the valuable ones where you get to really think and and map out for your users and then tell you know, go in and tell GLM 4.7 built into OpenCode to write in any of the end workflow that does this, and it can do it. Right? And and that's the level that we're at. So right now, I'm using for sure, I'm using OpenCode, ClaudeCode as my primary AI code writing tools.
I'm using any then as my testing platform, but eventually, you just kind of turn that into language in the Python scripts or or whatever it may be. And and I'm using things like super base or Redis or or what or, you know, just a postgres table and and wherever you store it as if you're back in database structures. But and Claude knows all that. And and GLM and and well, GLM through Claude code or or however you configure that. It it knows.
Scott (Better Brand)Yeah. So you've got a you've got a Twitch stream, right, where you're where you're doing this live you're live streaming this, and anybody can yeah. You anybody can follow your your open source decentralized AI Twitch and follow along right with you. Right? Well,
Kyle (Bow Tied Blue Fin)hopefully, my ops sec is good enough that I don't inadvertently have a Twitch stream going on to some people. But no. May maybe that's something that I should get started in and and tape my way through it to show some people how to build a simple app. But may maybe that's for the next couple of weeks.
David JohnstonLove it. That would be awesome.
Bowen BelmerShout out to the ideas guys. They're so dangerous now that you don't have to know anything.
Kyle (Bow Tied Blue Fin)Yes. Ideas guys who who have become engineers' worst nightmares, though, on on or infrastructure engineers' worst nightmares.
Bowen BelmerYeah. Yeah. Where's Alan? He's he's punching the wall right now. Yeah. Eric. Eric, we'd love to hear, though from from Morpheus Asia as well. I know yeah. Like David said, I know it's super late on your side.
Eric ChanYeah. Yeah. Hi. Good morning. Good afternoon, I guess. Yeah. So happened that it is I'm I'm still awake, so I joined the calls. So I I dropped into a say that I quite agree that we we are hopefully, it's keep highlighting on the privacy. But I I will say that we will like, privacy is important when the moments that the user, they hit some hit some issue or happen on that to to in the privacy regarding the privacy issue, then they will feel that this is important.
So unless that the the people will don't care, for example, like, they will feel that this lock is is safe until they get someone who has to hijack their lock. Then they'll need the or they I need to change the better locks. This this is the this is a very normal concept. This is this is why I think that we need to try to see from from the Morpheus side. We need to see that how to emphasize or how to bring up this this awareness that or seek out the urgency that there needs to be privacy. So this part is quite quite crucial on that. And this beside that, that this this all has been mentioned, Carlos, and we mentioned about actually ideas
turned into the turned into the so called development of using the AI to change. Right? This is quite quite true and what will happen now. Because these few days, I have been beat up a few couples of the developer that which is still not really a developer, but somehow they are starting to build something on AI, and which is I'm trying to bring them onboard to say that they try to use it use our office AI as the so called as a user to be resources
that to to to proceed the so called the because every I I found that all developer in the AI product, the main number one key killing things is killing them is is the GPU upgrade. We call this GPU upgrade as we call the term then. Because when it come to the AI product, most of the product that they they our application, they come out. They need to give people, like, for example, the onboarding need to have a free. After that, only the we upgrade into the professional or the pro version. That's why the free is actually is the biggest upgrade.
So from my side, actually, I also keep talking with them while we were trying to do a support with them, hand to hand support so that while you can you can try today or next to me, you can start to migrate into into the using more case API. And we will support you on this part to to build further more on the application or whatever. So so I see that most of them, it's usually yeah. Just need to be innovative and also creative so that they can proceed the next. Because AI now is very, very powerful. They can code it easily.
So this is what what my what my agrees on that and so all the topics that I created from this to in this field, you base everything up with some developer. Yeah. These are and you could be some updates from my side. Eric,
Bowen Belmerdo you have a pulse on the regulatory environment up by you guys? I know that, obviously, Morpheus Asia spans across, you know, multiple countries, and so this is maybe a fairly loaded question. But we were talking earlier about how The UK right now is cracking down on on social media specifically and how that'll likely, not likely. It will bleed into things like AI tooling. I mean, I was curious, like, just kinda what what the vibes are in Asia right now from your guys' perspective with, like, regulatory environment around AI. Yeah. Yes. Recently, because there's a crop, there's a there's a issue of the so called they are starting to generate the newism
Eric Chanfor photography or photography. Then that's why Malaysia actually, Malaysia I think Malaysia, Thailand, they actually start blocking the crops it's using in Malaysia and Thailand. That's the that's the scene. That's what happened last last week.
Bowen BelmerSorry, Eric. Can you say that one more time? You said that they're they're blocking image generation?
Eric ChanNo. It's blocking the crops. We are no longer able to use grog in Malaysia because of the Oh, really? Pornography happened that that they are generating their someone using their face and putting into the view. That that's that's why you can block it. Totally blocked. Totally blocked. Yeah.
Bowen BelmerWell, I don't want to I don't wanna promote spoof pornography, but I would imagine you can access Grock through Venice or Yeah. Or Morpheus.
Eric ChanYeah. Agree. If someone really wanted to, but ideally not for that use case. Yeah. Exactly. That that's the reason why I also talked to the developer that they they also some they are using your Grok. That's why I see that you are it's a good chance that you'll come you come over to use our our so called resources.
Bowen BelmerInteresting. Yeah. I am you know, from a a western perspective, that that was kind of why I assumed that, like, parts of Asia are likely one of the best breeding grounds for kind of the killer DAI applications for a lot of the reasons that we were talking about before where even beyond them making sense economically without considering the privacy and and permissionless nature of the tools. When you add those factors in, like, it is orders of magnitude,
higher in terms of opportunity. And, like, again, I don't have this vast knowledge of geopolitics in Asia, especially as it pertains to, like, you know, domestic issues in in some of these countries. But from a Western perspective, it tends to be tends to seem like the central controls around the Internet and access to applications and things like this are are often under more scrutiny scrutiny, let's say. Would you would you say that that's true?
Eric ChanYeah. Yeah. I guess.
Kyle (Bow Tied Blue Fin)I I would say that the data tells the story as well. Right? I was just looking at a map, this morning that showed DeepSeek, you know, even back to DeepSeek r one, which is not the best model right now. Like, penetration across the globe, and it has massive market penetration still in, you know, in countries that do have content restrictions, whether that's blocking, g p p, grok, or the others. You see that massive adoption in large
open source models. I do expect the the model selection to change to the GLM and the QEMs that that seem to be at the top of the market right now, but it's it's another tell that DAI is going to be used for many different reasons, you know, aside from privacy, access alone, and and being able to have that permissionless nature is is going to be a key one for many parts of the world outside of The United States.
Bowen BelmerYeah. Absolutely. It's so frustrating to me that I mean, this is the case with pretty much all technologies that provide any level of anonymity, but, like, the early adopters are often, like, morally not great actors because, like, those are the people who are, like, scratching and clawing for, basically privacy or anonymity to do the the thing that they shouldn't be doing in the first place. So, yeah, I I can't wait to see the, like
I don't know what the word I'm looking for is, but, like, productive applications and agents, like, coming coming out to try to win the the hearts and minds of of the average person who doesn't, like, follow, you know, deep tech development. Because, yeah, that's that's a bummer that, like, not only is Grok banned in in Malaysia and Thailand, but, like, the reason for it is just horrible for publicity, obviously. I mean, Bitcoin went through this too. Right? Though, I mean, Bitcoin was just, you know,
Silk Road money until it wasn't. And now, you have JPMorgan and Morgan Stanley and BlackRock and all these other entities allocating to it. So in due time, my friends.
Eric ChanYes. In due time.
Kyle (Bow Tied Blue Fin)But, yes. So thanks, Eric. That was really appreciated, and I think we had a a great update this week. Looking forward to next week, we'll either have another external builder or we'll have some people are building like crazy within the Northeast community. So we, you know, we love these types of of conversations to kinda get the get everyone's brain working on what can we build next and how do we build on top of Morpheus. And thank you guys for for attending.
Scott (Better Brand)Yeah. I'm I'm looking forward to you guys. I think two things. One is your AI workbench that you put in quite a list together there. So I look forward to seeing that in your your your inaugural video of your build. And then and then, Thomas, sounds like you have something or two up your sleeve that you'll be able to to to share, and we'd love to see what you you got building. So I appreciate you guys hopping on. David?
David JohnstonYeah. And Kyle, great article, by the way, about the need for agentic tools. It was really, really well written. I tweeted out sort of my thoughts. I think one of your paragraphs in particular sort of, you know, was was really insightful. So appreciate the thought leadership. It's nice to have, you know, folks like yourself writing, you know, effectively describing the vision, right, for where Morpheus can go, and I think that's that's really critical. So as somebody's written a lot of articles, I I I appreciate you jumping in and, you know, taking that baton and and adding to it, especially at your technical depth.
Kyle (Bow Tied Blue Fin)Really appreciate that, Dave.
David JohnstonCool. Alright, guys. Well, awesome session. Thanks for everybody jumping on the space, and we'll do it again next week and talk to more builders. Thanks, everybody.
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...
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...
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...
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...
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...