Tommy Eastman: Own Your Intelligence Stack
Transcript — TFTC
Article: https://www.tftc.io/hermes-agent-ai-open-source-tommy-eastman
Transcript page: https://www.tftc.io/hermes-agent-ai-open-source-tommy-eastman-transcript
Published: 2026-09-28
Machine transcription, lightly cleaned; may contain errors.
========================================================================

[0:07] Tommy Eastman: You've had a dynamic where money's become freer than free. Let me talk about a Fed just gone nuts, all, all the central banks going nuts. So it's all acting like safe haven. I believe that in a world where central bankers are tripping over themselves to devalue their currency, Bitcoin wins.

[0:25] Marty Bent: In the world of fiat currencies, Bitcoin is the victor. I mean, that's part of the bull case for Bitcoin. If you're not paying attention, you probably should be.

[0:34] Tommy Eastman: Probably should be.

[0:35] Marty Bent: Probably should be. I've been, uh, I've been very much looking forward to that. I was very excited when we first talked a couple weeks ago because I've been using, uh, your product for 4 months now and it's changed my life. You saw me when you walked in here. You thought I was on the phone. I was talking to my Hermes agent.

[0:52] Tommy Eastman: Talking to your Hermes agent. Better than talking to somebody, some would say.

[0:56] Marty Bent: I like the human, I like the human element. Okay. I do like talking to my wife, Greg, who's off camera here. He's okay to talk to every once in a while. But, uh, it's, uh, it's changed how I work pretty, pretty quickly. I mean, I'm 35. I've been doing TFTC for almost a decade now, and within 6 months, my whole workflow has changed. And Hermes is a critical part of that.

[1:21] Tommy Eastman: Yeah.

[1:22] Marty Bent: What— let's take a step back and talk about you personally, how you ended up at Noose Research, which built the Hermes agent. Um, you were in the Bitcoin industry, the crypto industry at Foundry before this, and that's— I mean, that's a whole nother conversation. I think it's really interesting how a lot of these up-and-coming AI companies are seeded with people that have been in the Bitcoin and crypto industry before.

[1:48] Tommy Eastman: Yep.

[1:48] Marty Bent: But how did you end up at Noose? What drew you to it? And what have you been doing there since you got there?

[1:56] Tommy Eastman: Yeah, so I was at Foundry, touched the Bitcoin stuff a fair bit, but was really focused on distributed and decentralized AI there. I just thought that AI, obviously an incredibly powerful tool, very important that we get the ideological part of AI shaped correctly. That was my first real exposure to that thinking about the importance of AI for people, for the individual, the sovereignty that AI enables. I didn't really like a lot of the kind of first manifestations of that, like the first projects that kind of tried to champion that. I wasn't that impressed with. I found Noose eventually. They were focused on post-training open models, um, to make them more pliable, more human-like.

[4:06] Marty Bent: So let's take a step back and basically explain Hermes. Let's focus on Hermes specifically. And let's, let's, because we've had this conversation off air a couple of times now, but speaking to the person who thinks that the crème de la crème of their interaction with AI is opening ChatGPT and hitting chat and and thinking that they're getting an experience. What does Hermes bring to the table and how should people view the Hermes harness and how to utilize it?

[4:37] Tommy Eastman: Yeah, well, that's the incredible piece, right? Is that we're so early. We feel like AI is everywhere and everybody's talking about AI, but there's only a small slice of people actually using AI. And then the vast majority of that slice of people are just using AI basically as advanced Google, right? Google on steroids. Like my mom loves Perplexity. She uses Perplexity all the time, but she's basically just, she likes it 'cause it's Google, but a little nicer. So where you start, where you make that jump from AI chatbot to agentic AI is when you start to enable actual autonomy. So these agents start to actually be able to make some decisions for you and carry out tasks without you having to sit there and monitor them 24/7.

[7:39] Marty Bent: Right.

[7:40] Tommy Eastman: The third piece of why Hermes Agent, I think, is so beautiful, and this is like the ideological holy grail, right, is it's able to be fully model agnostic. And that is important for a lot of reasons. And we've thought it's been important for a really long time at Noose. You've seen more and more attention come to this issue now, though, of data sovereignty. and provider lock-in. Nobody wants to be locked into a single big lab. Nobody wants these labs training on their data. People want to own that stuff because that's the secret sauce, right? That's the sauce of my company is this data. And if I'm sending all of that to a lab that's going to use that to train in the next model, well, I'm giving away the secret codes.

[9:13] Marty Bent: And it's, I mean, we, the, so my Hermes story, my name's Marty Bent, I'm a Hermes user. So I, OpenClaw, Over Christmas break, people are like, OpenCLaw. I'm like, one, it snowed like 30 inches in Philly one Sunday and the kids were watching a movie. I was like, all right, I'll try this out. So like first time I ever set up a, uh, VM and set up a cloud server, set up the Hetzner server, like went through this YouTube tutorial, set up OpenCLaw and connected it to Anthropic at the time when you could OAuth in before they, uh, before they prevented that from happening. And was immediately blown away, like, holy crap, this thing is insane.

[9:56] Marty Bent: Mm-hmm. So for about 4 months, I was building out that OpenCLAW instance and integrating more of our workflows into the agent setup and the harness in the cloud and like seeding the file system in the cloud with everything that I wanted this computer to be able to access. Along the way, like April, May, um, I was talking about OpenClaw and I heard people, uh, it was probably actually March, April, people beginning to be like, hey, you're using OpenClaw, you should like try out Hermes. And then like another similar situation where it was just a Sunday afternoon and, uh, I was sitting there and I was like, okay, like I've had 10 people like reply to some of my tweets, like you should try Hermes. And I was like, okay.

[10:41] Tommy Eastman: We love our community.

[10:42] Marty Bent: Literally went to my OpenClaw agent on Telegram. I was like, hey, Spin up a Hermes agent in your server and let me know when that's finished. And then I hooked up Telegram. And then within like 24 hours, I was like, okay, Hermes agent, shut down the OpenClaw agent. And we're gonna begin working. And it's been insane what we've been able to do, what I've been able to do on the team at TFTC over the last 4 or 5 months, just really seeding. the Hermes agent. But to your point, I think that's what, um, I really want to get through to people when we discussed this on our call a couple weeks ago is I view— I'm interested to get your thoughts on how you view this.

[12:07] Marty Bent: Mm-hmm. And I'm hitting /model changing for different task when I don't want to burn tokens or burn expensive tokens, or if I want to use a more private open-source model. I don't think most people understand. Like, I think we're at this— it's not that I don't think, I know most people don't understand that this sort of 3-pronged approach and what is the value, what is the relative value of the different prongs within that 3-pronged approach. And I think that's where the opportunity lies and why I love Hermes. Because I think you mentioned like the self-referential learning is part of the memory system. And I think there's a ton of people out there saying like, all right, AI's here.

[13:10] Tommy Eastman: Yeah, yeah, I think you're completely right. There's so much value. The returns are compounding. The quality of intelligence is compounding the more and more as the agent gets more and more intelligence, right? So that file system is incredibly valuable. The more The agents are only as good as the data that they are able to access, right? There's no— it's not some magic. And so the ability to give your agent access to as much context as possible and basically just let it go to work is the secret sauce. That's the real value in these systems. And I think it's almost hard to articulate sometimes to enterprises, like, Why would I use Hermes Agent? It's like, well, you can use it for everything. And that can— that's— it's like, it's almost too abstract, right?

[14:07] Marty Bent: Or too good to be true.

[14:08] Tommy Eastman: Yeah, yeah, yeah. It's like, okay, what does that mean? And it's like, the reality is though, you really can. And you've seen it, I'm sure. You really can use it for anything. Any monotonous task you may have, um, anything you want done, just throw it at your Hermes Agent and see what happens, right? I think the The low-hanging fruit where a lot of people immediately say, wow, this is incredible, is working with big data. If you have, whether it's your Stripe data, your customer data, the ability to create, to organize that, to create visualizations, to do modeling is stuff that these agentic systems excel at, right?

[15:35] Marty Bent: But they're becoming commoditized. That's where I'm—

[15:38] Tommy Eastman: Oh, in the long tail, for sure.

[15:40] Marty Bent: I think in the long tail, for sure. Over time, the ability to be model agnostic is only going to increase.

[15:45] Tommy Eastman: Yes. Yeah, I mean—

[15:46] Marty Bent: Because they're reaching parity.

[15:47] Tommy Eastman: 9 months ago, you couldn't use an open model for anything. Like, they were basically worthless, right? Um, and now you— I like to say, I think about, you know, 90% of users can do 90% of their work with open models, right? Like, there is definitely stuff that you want the, the biggest guns for, but tons of work can go to open models. And yes, that gap is continuing to close, and it's only going to become more and more commoditized.

[16:15] Marty Bent: Yeah. And at what point do, like, do you, like, what is the, uh, the minimum AI IQ level that you need to do all the tasks? Like, do you really, like, say a year from now, are you gonna need a frontier model to do your modeling or your CRM?

[16:30] Tommy Eastman: Yeah, the vast majority of people are never gonna do tasks that require these top-tier models. It's just not part of what they do.

[16:37] Marty Bent: No. And the, So let's take a step back and talk about AI broadly. Like, where do you think we are? What trends are you most interested right now? Is it the data center compute side? Is it this model race? Is it the harness? What are you most interested right now in the world of AI outside of Hermes and Noose?

[17:01] Tommy Eastman: Yeah, I mean, I think it's really interesting to follow how the open-source models have changed the market, right? Like I said, 9 months ago, you really couldn't use open-source AI. Now, if we transport back to January, February, you saw a bunch of prominent CEOs talking about, oh, our employees need a token max, token max, token max, like unlimited spend. Very quickly, they were like, whoa, way too much money, and want to find cheaper solutions, right? And that timed well with open models reaching the point where they can be these cheaper solutions. I think there's a lot of stigma around them for some people still about being, you know, mostly Chinese models, right?

[18:55] Marty Bent: No, and I think I just discussed this with Jordy Visser, who was on before you rolled in, but I think the Latham& Watkins story from a couple weeks ago, the law firm basically said, we're not using Anthropic or OpenAI, we're gonna buy our own GPUs and self-host our model. Yeah. I mean, that's a massive signal. Not only the models are good enough to do that, but also I think a trend that many people are not appreciating enough. And I think Alex Karp's been beating the drum about this, but like particularly if you're enterprise the size of Latham& Watkins or any Fortune 500 company, like the feeding of your company IP to Anthropic and OpenAI is, I think it's gonna get to a point where shareholders or boards of directors are gonna say, this is like a breach of fiduciary responsibility. Like you're just handing over the secret sauce to—

[19:49] Tommy Eastman: Yeah.

[19:49] Marty Bent: These model providers who are proving or who are literally launching products that they're training off of companies that are doing this, like that they're going to try to compete with you and cannibalize your business while it's providing you value in the interim. And so to your point about open models, like I think there will be a massive inflow of usage for increased usage of these open models from like a fiduciary perspective just to preserve your IP. Yeah.

[20:18] Tommy Eastman: Yeah, I mean, people are just throwing their trade secrets. It's actually a really interesting psychological phenomenon, right? Like, yeah, businesses are doing it, but also the average user, the ChatGPT user, the Anthropic user, people share so, so much with these models and almost like completely unaware of privacy or how that obviously could backfire and the different ends that that could go to. It's happening at the— at the individual level. It is happening at the business level as well. But I think at least now, more and more businesses are very aware of that. That's something that really resonates, um, with businesses that we talk to about, about Hermes Agent, is they want to— ever since that Alex Karp, um, spiel he went on, businesses want to own this. They're very, very aware. Um, they're very, very concerned about where their data is going. I think the general public is lagging behind, but I think there'll be a rude awakening for them as well.

[21:14] Marty Bent: Agreed. Agreed. What's up, freaks? This script was brought to you by our good friends, the Bitcoin team at Block. Bitcoin at Block. Bitcoin was originally introduced as a peer-to-peer electronic cash system. We all know that. The idea wasn't just to buy and hold, it's to use Bitcoin in many different facets. Holding is one facet, spending, receiving is another facet. We've been working towards this for 17 years and Block has put together a whole ecosystem of companies to really bring about the reality of Bitcoin as everyday money. So with Cash App, you can buy, spend, send, receive, earn, get paid in Bitcoin. When you're trading over $2,000, you're not going to pay any fees on that.

[22:49] Marty Bent: They own the machines that make it. Here's how it works. You buy the miners at Simple Mining. Simple Mining hosts them, runs them at their facilities in Iowa on industrial power rates you can never get on your own. The Bitcoin they mine goes straight from your machine to your wallet. 2 reasons I find this interesting right now. The second one is the one that nobody's talking about. One is the economics, cheap power. This lets you convert your electricity into Bitcoin at a discount to spot. You're acquiring the asset below the sticker price everyone else pays. 2, and listen closely, if you've had a big income year, making a lot of money out there, I know a lot of you freaks are making a lot of money.

[24:02] Marty Bent: It's funny, I was taking the Meta Muse agent for a spin this morning. I was like, wow, this thing's powerful. But then how much of this information do I want to give to Meta and Zuck? And I think that's a question that a lot of people are going to have to grapple with. But I mean, history shows, particularly in the digital age, most people don't care. That's why I think a focus on enterprise where you have to care Eric, because money and shareholder value, for lack of a better term, is on, on the line. Um, but then for the individual, like running individual agents, like I think, uh, hopefully, uh, team at Noose Research and others can get the, the UX, uh, to parity with something like a Muse agent where it just works out of the box, which I think to your point, like that was one of the most impressive things.

[24:50] Tommy Eastman: Yep. Yeah, I think that, um, it's really important to make the product easier and easier for people to use, right? Because if it's, you know, hours of setup versus instant setup and the privacy is the only differentiator, that's, that's been proven time and time again that, that that's not enough. I do think on the individual privacy aspect, I think It probably needs some sort of massive event to catalyze people growing aware of this, of this concern. But I do think we're at a stage where, yes, it was bad when you shared all of your information in this digital age previously, but now the ability for agents to access it, to compile it, access it, spread it, In seconds, right?

[26:03] Marty Bent: Wasn't there the ability to like Google's like if you were sharing The, uh, the output and you create like a share link that was getting cached in Google search. Yeah.

[26:12] Tommy Eastman: Yes, yes, yes.

[26:13] Marty Bent: Something similar happened to Meta as well. Um, yeah, where they thought they were like having a private conversation with the AI, but they're posting it publicly. Uh, yeah, it's horrifying.

[26:22] Tommy Eastman: Horrifying, horrifying.

[26:23] Marty Bent: Yeah. And it's— so how would you just like— that's what I'm— I've been trying to discern because things are changing so quickly. Like last year at this time, like Opus 4.6 wasn't even out yet. Correct. So you didn't have, like, there was that Frontier model sort of breakthrough moment where it was like, oh wow, now this stuff can like really cook. And then a month later it was like, they were working on Open Claw. You guys have been working on Hermes, but like people didn't become aware of like this harness thing until that happened. So we're only like 9 months into this. What, where does it, like, how do you think this looks? Uh, even 3 months from now, a year from now? And what trends should people be paying attention to, particularly around user experience and workflows that they can incorporate using this stuff?

[27:11] Tommy Eastman: Yeah, well, what you said is exactly right, right? Like, at this time last year, Opus 4.6 wasn't out. I think the best way to describe that change, that clear step function increase in AI capability was, At this time last year, like, my top— the top 10 developers I knew, 0% of their code was written with AI. And it wasn't better than them. Too many issues, too many hallucinations, no ability to do long-running tasks. Now, 100% of their code is written with AI, right? Almost nobody— there's almost no reason to be writing code except with AI now. And people are doing incredibly long-running tasks with agents. And that was less than a year ago, like you said, like 9 months ago that change happened. And so to think about where we are from the next 9 months, it's really, really hard to be predictive. I think that— I think what's very clear is we're early on the adoption of AI.

[28:17] Marty Bent: Mm-hmm.

[28:18] Tommy Eastman: And we're going to start to see like just those basic levels, like engineers all using AI. We're going to start to see that all be integrated into Every enterprise. This year from this time now, this time next year, I think every enterprise, all code is written by AI, right? There's really no— there will be really no excuse at that point. I think it will be abundantly clear that you have to do this to keep up, to ship, because the companies that choose to do it are going to— it won't happen overnight, but it will happen fairly quickly, will drastically outperform and be able to cut expenses.

[30:54] Marty Bent: I'm a massive flight, hotel, train procrastinator.

[31:00] Tommy Eastman: Yeah, same. It's horrible.

[31:01] Marty Bent: I've got to go to— I've got to go somewhere in a couple of weeks and I should have booked the flight months ago. Yeah, I booked it yesterday. I was like, I'm paying the premium for this because it's been— and it's been on my calendar. And it's, uh, what, what are you seeing inter— like, so this is one thing selfishly I wanna get out. Like, obviously you interact with a ton of individuals and enterprises leveraging Hermes Agent. What are some of the most creative or mind-blowing ways or workflows that people have incorporated using Hermes?

[31:32] Tommy Eastman: Yeah, it's a good question. Um, or who—

[31:35] Marty Bent: or another way to frame it is like, who's utilizing it the best and why is the way they're utilizing it the best way?

[31:40] Tommy Eastman: Yeah, well, I think Noose Research utilizes it the best. No, but, but in all seriousness, we— I think we— it's really, really important for us, right, if we're gonna deliver this product to enterprise, that we are the number one power user of it. Um, and so we use it, we use it for everything. Um, I couldn't even try to quantify the productivity increase that Hermes Agent has applied to Noose. But it's funny when people ask us how many, you know, how big is the team, and we say, you know, 30 people, 35 people, they're often blown away just because if you've watched the rate that we've shipped on the Hermes Agent repo and the cloud products and the enterprise offering, it's something that would, you know, take an order of magnitude more people in the past.

[32:53] Marty Bent: Very aesthetically pleasing.

[32:54] Tommy Eastman: The aesthetic, we want things to look pretty. Stuff we put out, all of that is generated in Hermes agent pipeline. It's, it's very reproducible. It allows our designers to have a ton of, um, a ton of control over the end product, and it takes a ton of, you know, manual, you know, graphic design work and labor off of their plate. Um, one of the most interesting tools we have is, uh, an agent internally that all our employees can use And it basically has a bunch of key data sources of ours. So if a user complains or if a customer complains about a billing issue or, oh, something isn't working correctly, a news portal or my Hermes agent, there's some bug, we can just ask our internal Hermes agent.

[34:32] Marty Bent: Yeah, that's what I'm wondering. Like, I would love, again selfishly, to like, like give your team access to my Hermes Agents. Like, where am I fucking up? Like, because that's what I'm like continuously like trying to figure out is, all right, how can I better utilize this? Like, I always feel like I'm not utilizing just in AI generally, it's like I'm not utilizing everything to the best of its ability. And it's trying to like explore where the boundaries and the edges are. And I feel like every week I have an aha. I'm like, holy crap. Like I just, like one example, this will be the second time you're hearing this story in 2 consecutive episodes if you listen to the Jordy Visser episode, but I'll tell it again.

[36:38] Tommy Eastman: It's incredible.

[36:39] Marty Bent: Within 5 minutes.

[36:40] Tommy Eastman: It's incredible.

[36:41] Marty Bent: Like I'm at the bar having a beer and getting work done. Yeah. For a media company, it's like, hey, that is work.

[36:47] Tommy Eastman: We're getting— Yeah, well, no, right? I mean, that's a non-negligible amount of labor that you would have had to do. It's kind of a pain and a tedious task. And it's like death by a million paper cuts when you have to do all those things all the time and the amount of time that it frees up by you being able to just dump all of that on it is— I mean, how much— I guess I'd ask you, can you quantify the impact on business that, you know, Hermès Agent has had for, for you all over the last 4 months?

[37:17] Marty Bent: Uh, yeah, I mean, I think from— it's increased our profit margins, increased our revenue. Uh, one thing I've hated historically, and I've told this to people many times. Like, like, we're a small shop. It's me. I ran TFTC by myself for 6, 7 years, and it's only within like last 3 or 4 years that we've brought on any employees. I mean, there's 5 of us and some contractors. Uh, I did all— for many years I did like the recording, the editing, the publishing. Uh, but the bane of my existence, and as funny as what I did at Barstool Sports when I worked there, was ad sales.

[38:48] Tommy Eastman: Yep.

[38:48] Marty Bent: Our demo data, like our, our stats on YouTube and whatever, and like come up with a deck 5 minutes later. Uh, it's got an HTML page on Vercel, beautiful deck that can actually sell. Like on top of that, it's like, let's make sure we're pricing this right. Like run market research about podcasts of our size. Like what is the CPM?

[39:07] Tommy Eastman: Mm-hmm.

[39:08] Marty Bent: Is there any premium on our audience specifically? Like go do that research and come up with like accurate pricing. And that has led to more closed ad deals for us. So yes, it has. Long way of saying yes, it has.

[39:21] Tommy Eastman: Yeah, but I like what you said too about, um, you know, you have a hard time personally manifesting aesthetically pleasing stuff in the digital age, but Hermes Agent enables you to do that.

[39:31] Marty Bent: Yeah, right.

[39:32] Tommy Eastman: And that's what's so cool about it is if you're, if you're a high agency, if you're a high agency person right now, there's no better time in the history of the world to exist, right? Um, Like, I can't write code anymore, but I can, I manage our compute clusters and I can get all of the visualizations, all the telemetry that I could possibly want from it just by interfacing in natural language, right? And build dashboards that help inform our business decisions and allow us to see how we're acquiring new users and how our compute cluster's health is. And that would take data scientists, front-end engineers, it would take a series of people to do that in the past. And now I'm, just prompting Hermes Agent and it's building out what a full team would have to build historically. So, it empowers you to do pretty much anything in any dimension that you want, in any domain that you want. I mean, it's unbelievable.

[40:25] Marty Bent: Freaks, look at me. I'm glowing. I've got an angel's halo going around me. You know why that is? Is I feel good. I feel taken care of. I feel blessed, healthy, happy. And, that is because I'm a CrowdHealth member. My family and I have been CrowdHealth members for 5 years now, literally this month, 5 years ago we joined CrowdHealth. We've had 2 babies, we've had multiple health events, and we're never going back to health insurance. CrowdHealth is crowdfunded healthcare. So you sign up for CrowdHealth, you pay a monthly fee, you help out with other people's bills, and it's significantly cheaper than health insurance. We were on COBRA as a family of 3 when I left my last job before I went full-time to CFTC, went on CrowdHealth, Now, as a family of 5, we pay, I believe, $700 a month. It's significantly cheaper. They're going to negotiate prices lower for you. They've consistently negotiated healthcare prices as much as 50%, 60%, 80% in many cases.

[41:18] Marty Bent: They help out with babies. If you have a pregnancy, you pay the first $3,000 and the crowd covers the rest. If you have a regular health event, you pay $500 and the crowd pays the rest. com, sign up today, use the code TFTC, opt out of health insurance. I'm uninsured, baby, and I love it. com and you'll get $99 a month for the first 3 months that you're on the CrowdHealth platform in the community. Bitcoiners, you found sovereign money. Now find sovereign health and sovereign healthcare. All right, freaks. You know I don't take sponsored money from products I wouldn't use myself. So listen up. The Aven Bitcoin Visa card is one of the most interesting things I've seen in the Bitcoin lending space in a long time.

[42:17] Marty Bent: 10 years. That's 10 times longer than most lenders out there, or go interest only for up to 5 years. Rates start at 8.99% APR. For a product that lets you keep your stack and still access liquidity, it's hard to beat. On top of this, guess what? You also get 2% unlimited cash back every time you use the card. Spend fiat, keep your Bitcoin. That's the whole game. If you've been stacking for years and you need liquidity without triggering a taxable event, this is worth a serious look. Go to aven.com/bitcoin. That's aven.com/bitcoin. Check it out. I mean, let's shift to this, the compute side. What are you, what are you seeing on the compute? You're talking about your cluster. What, on the physical side of things, like you actually need the GPUs, you need to download it, like what are you seeing there? Like what is it like managing a cluster? Maybe we'll start there.

[43:05] Tommy Eastman: Oh, all sorts of headaches managing clusters and compute for sure. What I think is really interesting about the way compute has shifted, it's shifted with use, which is I guess obvious, right? But I think it's worth noting, again, we said this time last year Opus 4.6 wasn't out. developers weren't using it to write code. Good developers weren't using it to write code. I was using it to write code, but good developers were not using it to write code. At this time last year, you could, if you needed a cluster of whatever, 128 GPUs, I could send a text out and have 10 different providers bidding to win that. Now it's the inverse. I basically would have to go and and text 10, 20 different people and say, I need a cluster, I need a cluster, I need a cluster. Nope, we don't have any. We don't have any. We don't have any. It's unbelievable how squeezed the HPC industry is right now.

[44:05] Marty Bent: Yeah.

[44:06] Tommy Eastman: I think we're up against a lot of, you know, land and power limitations. I think you're seeing the regulatory environment has just really squashed the ability to continue to grow that. But also just the usage was such a parabolic increase in inference needs that we weren't prepared for it, right?

[44:30] Marty Bent: Yeah. And so to that point, who was getting access, obviously outside the frontier model? Because you have training, obviously you have the inference, and obviously Anthropic, OpenAI, Meta, Google, SpaceX, they all have incredible access to this stuff. But I think we've discussed like OpenRouter. I think Stripe's acquisition of OpenRouter is a very strong signal. And I think this NeoCloud sector is going to, I think basically having open market for different models is a very smart idea, especially if we believe that being model agnostic is going to be a trend that strengthens. In the future, to that point, like how do you get priority? Like how do you both on, okay, backup infrastructure side, what makes you a good cluster?

[45:49] Tommy Eastman: Yeah. So on the first piece, yes, it's managing these nodes is difficult. There's a lot of entropy in the management of these nodes that I don't really fully understand, but there's a lot of downtime. If you're talking about managed inference specifically, there's throughput and latency, throughput, latency, uptime. And so you, I mean, these providers do wanna work to have a good experience for the end user because it can be a huge headache and people are paying a lot of money for these nodes. And if their throughput isn't met, if their latency requirements isn't met, it really ruins the end user experience really, really quickly and people are gonna stop. people are going to stop paying for, paying for machines from a specific provider. Right now, the market, I think, is so squeezed that there's way more wiggle room and tolerance for that stuff just because it's like, where else are you going to go?

[46:47] Marty Bent: Mm-hmm.

[48:11] Marty Bent: And I'm happy we're discussing this now because I meant to bring this up earlier. I think it's important to clarify for people, we were talking about open-source models, many of which are predominantly trained, uh, in China and then released. Uh, people hear that, they're like, oh, I'm sending all my data to China now. That's not always the case. Like, if you're connecting directly to DeepSeek servers, like, yes, you are. But I think the understanding around, like, no, like, it's open source, so you can have American companies go buy GPUs and download the model there, and you're, you're not serving it to Chinese servers. It's a Chinese model run on American servers.

[48:47] Tommy Eastman: Correct. Massively different, right? And that's been a huge education piece that we're trying to drive home is You can have all of these cost savings. You can get really good performance at a fraction of the cost, 60%, 70%, 80%, 90% cheaper than US closed frontier labs. And you can do it with— yes, it's a Chinese model, but it's running on infrastructure owned by a US company in a US data center on US soil, right? And those are where your data flowing is 2 totally different things.

[49:16] Marty Bent: Yeah, well, you can say that people are like, it's phoning home to China though, right? It's like, no, not exactly.

[49:21] Tommy Eastman: Definitely not.

[49:21] Marty Bent: And then, uh, like the pro— again, going back and like tying in privacy to this too. This is what, uh, I know we were talking about Maple AI, uh, Venice has obviously exploded in popularity, uh, PPQ is another one. Uh, I am very bullish on running these open source models in trusted execution environments, just like to increase the privacy layer, because that Shout out to Mark Goodwin, who's been on the tip on Twitter saying like, even if you're using open source, it doesn't mean decentralized. There's still somebody hosting those servers at the end of the day. But with like trusted execution environments, being able to run the inference in a secure enclave and have everything end-to-end encrypted from your device that's baking the prompts to the input in transit is encrypted, then— You do everything on the secure enclave, you re-crypt it, send it back. Like, I think that's a massive opportunity too. I'm just rambling now.

[50:19] Tommy Eastman: Yeah, and I think that, and I think local models, local models as well. And local models are continuing to get better. Hardware, consumer hardware is continuing to get better and better, right? And so the ability for you to run a model on your local machine and do, again, it's not gonna be 100% of your work, but a good chunk of your work, especially if you have sensitive workloads, you can run those on local models and, and get, get real work done.

[50:45] Marty Bent: Well, this is actually something, I'm happy you brought this up. This is something I've been like thinking of, like, I'm not smart enough. I'm, maybe I'm not, I won't—

[50:53] Tommy Eastman: You're smart.

[50:54] Marty Bent: I won't denigrate myself. I'm smart, but I, I do not have the skills to post-train an open source model on my own data. That is like something if like Hermes, if Noose could figure out a way to take what I built with my Hermes agent and be like, okay, I wanna use this model, um, this local, this SLM, like locally, uh, Hermes AGI, I have all the context of what we've been working on over the last 6 months. Like, can you help me post-train this model on our specific data? Like that, I think, at least for my purposes, that's the one thing I've been, uh, somewhat obsessed with for the last week or two is like, okay, I've gotten to this point. Like, what else can I do? Yeah.

[51:30] Tommy Eastman: How can you squeeze more performance out of this?

[51:32] Marty Bent: How do we post-train a model on all of our stuff? Like, How do I do that? Yeah. That's what I'm looking for. If anybody wants to help me, let me know.

[51:38] Tommy Eastman: Yeah, we can do that. We can get you that. So that actually was part of the reason that Hermes Agent was created. It wasn't created to be this outward-facing product. It was created really out of like a completely utilitarian birth. Our co-founder Technium created this Hermes Agent to help as a small AI research lab, right? You're always trying to find Like these moonshot ideas to push you to the frontier. And Technium created this self-evolving Hermes agent to help us become frontier model trainers. So the original birth of Hermes agent was really to do— it was in part to do that. And that's something we should do a job of highlighting, how you can use Hermes agent in that stuff as well.

[52:26] Marty Bent: Yeah. So what, uh, what's the plan with Noose? Like, you have the agent, you have Noose Portal, you got this enterprise. What, uh, what are you guys striving for, um, in terms of what this company looks like? Um, and I know you said this on the phone, like, who knows what any company looks like. Like, what are your, what are your north stars and who are you building for at the end of the day?

[52:49] Tommy Eastman: Yeah, yeah. I think, um, who are we building for is, is everybody. And I think I answer that really intentionally that way because I think we've built this product in a way that it meets you wherever you are. The open-source repo, if you want to run a local model on your local network and never touch any Noose Research managed product, that's great. Like, the Hermes aging repo will always be able to empower you to do that, and we'll continue to improve it and maintain it. If you want certain things managed, like, great, we'll do all that as well. We will be the best— we are the best place to have your Hermes agent hosted and managed.

[55:30] Marty Bent: Yeah. Meet people on their device, that's been one of the interesting observations I've had over the course of the years. who attracts to what specific form factor. Like I was telling you, like I haven't used Claude Code or Codex Desktop in 7 months. People are like, how are you doing that? Like, are you trying to get left behind? I'm like, no, I literally, it's connected to all these different models. And my go-to form factor is just like voice-to-text in Telegram. I have a group chat with the agent, just the agent and me, but the group chats allow you to create topics. Yep. So we have topics that represent different sessions that are AdOps, Bitcoin Brief, or the commoner newsletter builder.

[56:40] Tommy Eastman: Yeah.

[57:32] Tommy Eastman: Yeah, it's a massive problem, right? And you're exactly right. There isn't some golden solution where, oh, everybody is going to use this as the way to interface, right? At least certainly not yet. And I think the stance that we generally take is don't be, Like, we, we're definitely opinionated on some of the best ways to use these things, but we don't wanna ever lock in unless it's blatantly obvious that that is the place that we need to lock in, right? So can we— or so we allow WhatsApp, we allow Telegram, we allow email. Like, wherever you want this thing to live and meet you is, is important 'cause people use it, people use it in different ways.

[58:25] Marty Bent: Yeah. No, it's funny, when we, like, as you're mentioning that, like email, the Hermes agent has an agent.toomail email address. And so like for my wife, whenever she's like, ah, like we're in the market for a new car, and she's like, ah, she's like doing it the old way, I'm like, hey, just email Martin. I gave him permission to email you back. Like tell him what you want and he'll go do like a search search for you. So like, my wife is emailing Martin, the more sophisticated Marty, is, uh, is doing all this work. But it's like, it's been fun watching my wife just, just, she'll be like, oh, I emailed Martin today and he like did some good work for me. I'm like, that's pretty incredible.

[59:04] Tommy Eastman: Does she like Martin more than you?

[59:05] Marty Bent: No, no, unfortunately. I hope not. Not yet, at least. Maybe at some point. Maybe when robotics get here. Shoot, maybe we need to put out— I'm kidding, I'm kidding. But no, it's just like the, yeah, like that's one form factor where it's like, hey honey, instead of like you taking my phone and getting Telegram, just email Martin. Yeah, yeah. And he'll get you a list of cars in the area that are within the range of what you're looking for.

[59:34] Tommy Eastman: Yeah.

[59:35] Marty Bent: She's just been going back and forth with him in an email thread.

[59:37] Tommy Eastman: Yeah, and it's probably a magical experience for her, right?

[59:40] Marty Bent: Yeah. Yeah. Yeah.

[59:41] Tommy Eastman: And awful, and awful, it's all of those monotonous tasks like going and searching and getting quotes from 6 different vendors and finding the car you want and the reviews, all of that is— it goes from these incredibly laborious processes to, you know, a few minutes texting your Hermes agent.

[59:57] Marty Bent: No, and then we have, uh, Ed on our team. He has his agent, uh, and our agents are emailing each other. Yeah, like, so if we have stuff in the business, like, Ed in his Telegram or his Hermes desktop app with his agent, like, all right, email Martin, um, and tell him that we did this thing in this part of the business and make sure he updates his second brain to recognize that. Yep. And I feel like, I still feel like I'm not utilizing it. Like, I'm like, we do that and something like a workflow like that will emerge. I'm like, is this the right way to do something like this?

[1:00:57] Tommy Eastman: Mm-hmm.

[1:00:57] Marty Bent: Or like the company is like essentially like a knowledge layer and your job is to give it the best, most robust knowledge that every employee can access. And so, I haven't let everybody into the group chat on Telegram. However, we do have like Discord and the agents run wild in Discord and they can talk there and interact with them. So that use case of creating a second brain for a company specifically and then getting access, the right access to individuals on the team is a massive problem. Not a problem, massive opportunity. Because when you think about like permissions, especially if you're like selling enterprises, like, okay, I want Becky from finance to have access to this data, but not this data. And like just building the guardrails for that.

[1:01:43] Tommy Eastman: Yep, totally.

[1:01:44] Marty Bent: Massive opportunity.

[1:01:45] Tommy Eastman: Yeah, it's a massive opportunity. And I think there's a bunch there too, as far as how can you use Hermes Agent to get more out of every employee, right? And I think what you see is certain employees, it's a skill to be able to prompt these agents, right? Like different people get different performance out of agents. It's like very much a feel thing. And I think you have, you have people that are incredibly, incredibly proficient, like your top performers. Um, Hermes Agent lets you bring other people up to their level, right? And that's what we're rolling out for enterprise is, is, is, you know, this concept of, of a collective wisdom, a collective brain where you can take performing employees, top skills and deliver those to relevant employees, and they immediately get boosts in their performance.

[1:03:14] Marty Bent: Yeah. What do you think this does for the job market? Are you, uh, the permanent underclasses as quickly quickly approaching or a world of a billion flowers are going to bloom?

[1:03:27] Tommy Eastman: I think we could have a whole other podcast on that. I think I land somewhere in between, right? I think that what this does is, you know, it enables you to not do all of these monotonous tasks, right? If you've— I think my big—

[1:03:47] Marty Bent: Oh my God, I'm going to lose my laptop job. Yeah.

[1:03:49] Tommy Eastman: My bigger concern is, oh, you Like, you're worried about not just like, you know, making Excel sheets for the rest of your life or whatever. Like, that's your—

[1:03:56] Marty Bent: I'm really worried about the paycheck. That's your concern.

[1:03:59] Tommy Eastman: Yeah, like, I think what this, what this should do is free people up to do, you know, more thoughtful and higher agency tasks than sitting there and like button pushing. Um, so yeah, I mean, is it going to be this straight shot to a million flowers? No. Like, there's of course going to be bumps in the job market as these disruptive technologies move through their natural growth and integration and flow. But I think if you— the optimistic view is that this will enable humans to do things that are way more fulfilling than what a lot of people have convinced themselves is foundational to their lives.

[1:04:38] Marty Bent: Yeah, that's sort of philosophical musing I've had over the last year. Because you mentioned earlier, like, if you're high agency, you can get a lot out of this. And that's the question I have. Is there just like a natural distribution of individuals who are inherently high agency and a very large percentage of which are not? Or is there an ability to engender high agency in people? And do AI tools inspire them to become high agency? I don't know. I think that's what we're going to find out in the next decade.

[1:05:13] Tommy Eastman: We'll find that out. I'm sure it's some mixture of both.

[1:05:16] Marty Bent: Yeah. What's your most contrarian take on what's going on right now?

[1:05:22] Tommy Eastman: My most contrarian take of what's going on right now. I don't know if it's contrarian. I'm sure it's— there's a lot of people I think that probably share this viewpoint. I never know how pigeonholed I am, or sorry, how echo chambered I am. But I think that if you want to talk about existential risk and the way that existential risk of AI, like AI, misaligned AI killing all humans or whatever, I actually sometimes believe in— I believe there's some degree of credibility to that argument as a possibility. But what I think is, I think where we get to crazy town really fast, and it's really, really disingenuous, is you see large companies and lobbyists pushing for a stop of open source, right?

[1:07:16] Marty Bent: Incredibly. It's almost disgusting to the point, because it's like, if— I don't think most of the public is aware because they don't pay close enough attention, but it's like very clear to me. It's like, okay, you're looking for the— you've raised a ton of capital You're deploying a ton of capital into this infrastructure build-out, particularly for the frontier labs. And you have a need to get a return on invested capital.

[1:07:44] Tommy Eastman: Mm-hmm.

[1:07:44] Marty Bent: And you're seeing these open-source models begin to nibble at your feet. Not only that, like many people are turning to them for 90% of the work for their individual use cases. And you're trying to shut, pull up the ladder and shut the door behind you. Just so you can save those profits. And I think when you consider how profound of a shift this technology is for humanity, the fact that there's literally 2 or 3 companies trying to pull up the ladder and control it all, it's just incredibly, I would say, like evil at the end of the day.

[1:08:16] Tommy Eastman: Well, where did all the data come in the first place?

[1:08:18] Marty Bent: Right?

[1:08:18] Tommy Eastman: Like it's all of our data that trained it.

[1:08:20] Marty Bent: I mean, the fact that they were buying like centuries-old books and like cutting off the bindings And then burning them. It's like, what the hell?

[1:08:27] Tommy Eastman: Yeah, it was all our data that it came from in the first place, right? And then they, yeah, it's just disingenuous.

[1:08:33] Marty Bent: Dealing with it with the New York Times right now, I believe.

[1:08:34] Tommy Eastman: Yeah.

[1:08:34] Marty Bent: I think that, I guess, copyright case is moving forward. Tommy, it's been a pleasure. We should do this more. I am gonna pick your brain off air about what I should be doing with my agent.

[1:08:49] Tommy Eastman: I love it.

[1:08:50] Marty Bent: Any final thoughts, final words of wisdom before we wrap up here?

[1:08:54] Tommy Eastman: I don't know about words of wisdom. I think at some point it will be very obvious that you should care about privacy in the age of AI and agentic AI. I think that the more proactive you can be about it and the steps that you can take to empower a world where you have more ownership as a business and as an individual of your AI and where the data that you give to your AI flows, the better served you'll be and you can avoid a lot of headache down the road. And so I think looking for obviously Hermes Agent, but looking for tools that enable you in the AI realm to have good controls over your data, but do it in a simple way that isn't super burdensome and a huge headache for you, I think will be a really, really valuable investment going forward.

[1:09:52] Marty Bent: I completely agree with that. Thank you for joining me. Thank you and the team at Noose for building Hermes. It's been incredibly powerful for us here at TFTC, and excited to see what you guys build next.

[1:10:05] Tommy Eastman: Love it. Thanks, Marty.

[1:10:07] Marty Bent: Thank you. Peace and love, freaks. Thank you for listening to this episode of TFTC. If you've made it this far, I imagine you got some value out of the episode. If so, please share it far and wide with your friends and family. We're looking to get the word out there. Also, wherever you're listening, whether that's YouTube, Apple, Spotify, make sure you like and subscribe to the show. And if you can leave a rating on the podcasting platforms, that goes a long way. Last but not least, if you want to get these episodes A day early and ad-free. Make sure you download the Fountain podcasting app. You can go to fountain.fm to find that. $5 a month gets you every episode a day early, ad-free. Helps the show, gives you incredible value. So please consider subscribing via Fountain as well. Thank you for your time, and until next time.