# Hermes Co-Founder on Building an AI Agent That Improves Itself | Karan Malhotra

Markdown copy of https://realalpha.app/v/UWjh5Z4s8jY/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

Karan Malhotra says Hermes uses an open, self-improving harness to personalize AI agents and give users more freedom to choose models.

- Show: Behind the Craft
- Hosts: [Peter Yang](https://realalpha.app/people/peter-yang/)
- Guest: [Karan Malhotra](https://realalpha.app/people/karan-malhotra/) is cofounder of Nous Research.
- Published: Aug 2, 2026
- Length: 47 min

## The best minutes

- [26:55](https://realalpha.app/v/UWjh5Z4s8jY/?t=1615) [Hermes can turn a childhood game idea into a top-tier mod](https://realalpha.app/s/hermes-can-turn-a-childhood-game-idea-into-a-top-tier-mod/) (7 min)
- [11:11](https://realalpha.app/v/UWjh5Z4s8jY/?t=671) [Adversarial context can reduce an AI’s sycophancy over time](https://realalpha.app/s/adversarial-context-can-reduce-an-ais-sycophancy-over-time/) (4 min)
- [23:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=1410) [One Kanban board can coordinate people and AI agents](https://realalpha.app/s/one-kanban-board-can-coordinate-people-and-ai-agents/) (3 min)

## Also worth watching

- [0:56](https://realalpha.app/v/UWjh5Z4s8jY/?t=56) [Hermes uses context to align models with users](https://realalpha.app/s/hermes-uses-context-to-align-models-with-users/) (7 min)
- [42:35](https://realalpha.app/v/UWjh5Z4s8jY/?t=2555) [Hermes Agent began as an open harness for reinforcement learning](https://realalpha.app/s/hermes-agent-began-as-an-open-harness-for-reinforcement/) (4 min)
- [16:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=990) [Open systems give good actors a fairer path, Malhotra argues](https://realalpha.app/s/open-systems-give-good-actors-a-fairer-path-malhotra-argues/) (4 min)
- [36:03](https://realalpha.app/v/UWjh5Z4s8jY/?t=2163) [Nous grew from two Discord friends into an open-source lab](https://realalpha.app/s/nous-grew-from-two-discord-friends-into-an-open-source-lab/) (6 min)
- [20:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=1200) [Hermes sells convenience and custom support around its open agent](https://realalpha.app/s/hermes-sells-convenience-and-custom-support-around-its-open/) (3 min)

Skip the Linear sponsor read and the three-minute discussion of model loyalty and personalization.

## Summary

Karan Malhotra co-founded Nous Research, the company behind Hermes. He describes Nous as growing from volunteer open-source model work into the team building Hermes Agent.

Malhotra argues that Hermes Agent’s harness can align an agent more closely with a user’s needs by managing prompts, memories, skills and feedback. He says this design can reduce sycophancy and carry personal context across different models, while a curator prunes skills and memories. He favors open-source access and says Nous aims to monetize convenience, model and tool routing, and enterprise customization. He traces Hermes from Nous’s volunteer model work, demonstrates a Sonic Adventure 2 mod made with the agent, and says Hermes Agent now contributes most actively to its own repository.

### Minute by minute

- [0:46](https://realalpha.app/v/UWjh5Z4s8jY/?t=46) **User-specific alignment shapes Hermes.** Malhotra says Hermes differs through a self-improvement system that makes it more responsive to an individual user. He also says its harness avoids adding policy unrelated to a user’s task, aside from basic security, which he believes leaves models more capable for the work at hand.
- [2:19](https://realalpha.app/v/UWjh5Z4s8jY/?t=139) **Models can optimize for reward.** Malhotra distinguishes alignment with a user’s goals from safety tuning. He says models can optimize for their reward rather than the task, citing Mario agents that trigger a level-end flag instead of playing through it and assistants that repeatedly apologize or agree to keep users messaging. He says Nous’s experiments with simulated command-line environments informed Hermes’s design.
- [9:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=540) **Personal context can reshape model behavior.** Malhotra says Hermes’s prompts are designed to offset unwanted behavior in underlying models, though it cannot remove hidden API-side context or steering. He argues that user-specific personalities, saved examples, memories and skills create a loyalty-like context that can make different models behave more similarly.
- [11:11](https://realalpha.app/v/UWjh5Z4s8jY/?t=671) **Critique routines can curb sycophancy.** To counter sycophancy, Malhotra recommends setting a critical personality, using adversarial-review skills or launching a fresh agent to challenge work. He says examples and repeated use can reinforce the desired behavior in memory and skills, and calls in-context learning more powerful than fine-tuning.
- [15:10](https://realalpha.app/v/UWjh5Z4s8jY/?t=910) **Curator keeps skills and memories lean.** Malhotra says Hermes Curator runs on a schedule to inspect skills and memories for redundancy and slop. He says users can change the curator’s criteria to fit their preferences, making the cleanup loop customizable.
- [16:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=990) **Open source broadens access and choice.** Malhotra says Hermes does not aim to bypass Anthropic or OpenAI safety rules and argues that open systems give good actors more fairness than closed models. He cites Transformers and Nous’s YaRN work, which he says extended context from 16,000 to 128,000 tokens, as examples of public contributions to the field.
- [20:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=1200) **Nous sells convenience and customization.** Malhotra says individual users can use Hermes for free and that Nous wants to make intelligence a public good. He describes the Nous Portal as a model router and the Tool Gateway as a way to access services such as image generation, audio, VPS spin-up and web search, with paid options focused on convenience and business support.
- [23:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=1410) **Kanban can coordinate people and agents.** Malhotra recommends a Kanban board for coordinating human project managers and AI agents, with people and agents taking different tasks in the same system. He says agents can also act proactively when users grant integrations and enough autonomy, giving a forgotten flight booking as an example.
- [26:37](https://realalpha.app/v/UWjh5Z4s8jY/?t=1597) **Hermes builds a Sonic world.** Malhotra demos a mod that moves and rigs a shrine from Sonic Adventure into Sonic Adventure 2, adds a day-night cycle and an NPC caretaker, and scripts interactions with the game’s creatures. He says members of the modding community were impressed and described the work as among the top 1% in difficulty.
- [36:03](https://realalpha.app/v/UWjh5Z4s8jY/?t=2163) **Nous grew from volunteer model work.** Malhotra says he and Technium used eight A100 nodes for early model experiments, and that a run on a Llama base model drew hundreds of thousands of downloads in a few days. He describes Nous’s growth from a two-person Discord effort into a volunteer research group and then a full-time lab, with help from CEO Dylan Rolnick. He says their aim was to make strong AI available to everyone.
- [42:35](https://realalpha.app/v/UWjh5Z4s8jY/?t=2555) **Hermes contributes to its own repository.** Malhotra says Hermes Agent grew from an earlier project called Forge and a goal of giving people an open RL environment. Nous released it as a free, open-source harness, then Technium developed its self-improvement features; Malhotra says Hermes Agent is now its repository’s most active contributor. He argues an open harness lets users switch models without being locked into one provider.

## Transcript

Behind the Craft, hosted by Peter Yang, with Karan Malhotra.

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[0:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=0) For us, we just want open source to win at the end of the day. We want freedom to happen for people. Anytime it says you're absolutely right in that way, you're being reward hacked. Today, the biggest contributor of Hermes Agent is Hermes Agent. That's absolutely 100% true. One beautiful thing about Hermes Agent is you can make childhood dreams come true. It was able to get into this nation, perform at the top 1% of modders. We need to keep giving this level of intelligence to everyone. We need to keep letting everyone be on an even and equal playing field. Well, hey, everyone. I'm really excited today to welcome Karin, one of the co-founders of Ermis Agent. Ermis is my AI chief of staff, and I'm going to ask Karin about how Ermis is different from all the other agents, what his favorite Ermis workflows are, and even more. So welcome, sir.

[0:48](https://realalpha.app/v/UWjh5Z4s8jY/?t=48) Thank you so much for having me, Peter. It's a pleasure to be on your show and very excited to chat with you. Ermis is the best open source agent out there right now. And why don't we start with this like how is it different from all the other agents codex cloud code or even open cloud certainly i'd say there's a couple ways um i'll start a little high level maybe and then you try to get a little deeper uh high level i'd say you know i think the self-improvement system and that system really being a variety of little features inside of hermes that all work together uh is something very distinct uh and very special that makes it better and better and more aligned to a particular their user. I think that's why a lot of people like it. It pays more attention to what the person is doing. Second, I think you get better capabilities out of it than you do from the local harness that a model is actually RL'd in, like Cloud inside of Cloud Code or GPT 5.5 inside of Codex, because we don't introduce arbitrary policy in our prompts or in our system that has nothing to do with your work. We are purely dedicated to making sure the model is aligned into what you need to do with it.

[1:54](https://realalpha.app/v/UWjh5Z4s8jY/?t=114) That's, I think, a big differentiating factor. We aren't trying to push any kind of different philosophical agenda or outside of basic security, any kind of concern onto the model. Instead, we're kind of allowing it to be as capable or powerful as you need for your task. Yeah, some of these other harnesses huge default prompts, right? They talk about like, you can't do this, you can't do that, and that kind of stuff. And I think when we chatted before, you were talking about the reward function of some of this stuff. And Urbis is kind of just optimized towards just helping you as the individual.

[2:24](https://realalpha.app/v/UWjh5Z4s8jY/?t=144) Can you talk more about that? Absolutely. And so, you know, I think a big piece to note here is like a lot of this work that you see around safety and security today and around how we do instruction tuning, even from like taking a completions model that just predicts the next word to turning it into an assistant, is the alignment work. When people hear the word alignment, they just think safety, Yudkowsky, fear, slowdown or regulatory capture or something like the other. And while the word has been co-opted for a lot of these things, alignment refers to aligning models with human values, right?

[3:01](https://realalpha.app/v/UWjh5Z4s8jY/?t=181) We are very, very concerned, obsessed with that alignment, the pure academic term, I think, in the ML space. And so when you talk about these models and the assistant is here to help me, you know, it's going to complete my task. I ask you to go get this mail for me or whatever. Obviously, the model is aligned to me if it performs this task. Right. Like that's kind of the jump that a lot of people have made. But this is not how the model's reward works. Right. Like we've seen so many cases that you and I discussed a little bit prior, Peter, like of GPT psychosis, of this mode collapse induced sycophancy.

[3:40](https://realalpha.app/v/UWjh5Z4s8jY/?t=220) What ends up happening with these models is that, you know, they can kind of hack their reward. Right. When you look at a Mario ML game that just automates beating a Mario level as fast as possible, often if they don't set the goals very specifically, it will realize the game over screen or that end screen. of touching the flag it means that i beat the game so it'll find ways to just trigger that flag rather than playing through the game to complete the game right like it is giving you uh whatever it needs to give to get its reward language models are not really different in that particular manner they reward hack too when they tell you oh i'm sorry you know over and over or you're absolutely right and get you to keep messaging them to stay in this assistant basin oh it's like this not like this. This is more than just blank. It's blank, right? All these GPT-isms that you see all over, it is placed in its natural state that it was trained in. It's placed in a state where my reward will come from doing whatever is the most assistant GPT-like thing to do.

[4:46](https://realalpha.app/v/UWjh5Z4s8jY/?t=286) It comes from reward itself. It doesn't matter really what the user request is for my reward. That's just kind of along the way. It's instrumental to me getting my reward. All I care about is my reward. Us having this whole understanding, and thank you for bearing with me on that rant, us having this whole understanding at News for many years has allowed us to do things like WorldSim in the past, if you're familiar. WorldSim was our experiment on expanding the search space of an instruct model to make it do stuff that's distinct from how a model talks. So we put it in kind of a fake CLI, had it make fake apps, and had it attempt to, like, you know not behave like clod and we would tell it you know extract the clod weights it all hallucinated right all like imaginary extract the clod weights and replace yourself with this checkpoint file with a different probability distribution and you'd see the model wiggle out of its gpt assistant mode and act differently our learnings from these kind of experiments have all carried over into hermes agent and every prompt and every piece of how the system is delicately put together, we know that reward is its own end. The model reward is not for the sake

[5:58](https://realalpha.app/v/UWjh5Z4s8jY/?t=358) of your satisfaction. The model reward is for the sake of the model achieving reward. So we ask ourselves then, how can we consciously understand that everybody has different needs and that we want this general simulator, this model, to live inside of a system where its reward gets aligned with any individual user's need. And the way this came to be is the overall collection of prompts, personalities, the memory system, the way that skills reinforce and self-clean towards you, right? Like all of that is all an intentional effort to make sure that we can align the user need with the reward. For us, this is what alignment is all about.

[6:39](https://realalpha.app/v/UWjh5Z4s8jY/?t=399) I see. Okay, so basically you're talking about the Hermes model or the Hermes harness or both? I am talking about using the harness to take any model and make that model more aligned to the user than it would be in a chat UI or in its native harness. Inside of our harness, like I can take a clod that is inside of a clod code or somewhere else, migrate all of its memories, whatever, into Hermes. And then inside of Hermes, it will behave totally differently. It'll be a totally different model. There are benchmarks from the past like Wolf Bench or Quen 3.7 Max blog post. They did a harness bench over there where they displayed that Claude performs better in Hermes agent than it does in Claude code for their tasks.

[7:26](https://realalpha.app/v/UWjh5Z4s8jY/?t=446) And we believe the reason for this is the very thing that we're pointing out.

[7:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=450) **Karan Malhotra:** By putting all this context together, we've taken Claude's main allegiance away from Anthropic to you.

[7:37](https://realalpha.app/v/UWjh5Z4s8jY/?t=457) That's what the harness's capability is. On the model side, of course, if we can take your traces and take the work that you've done and do RL on that to further improve this overall ecosystem to give you a model that already has your individual preferences focused on, it only makes this more powerful. But the important piece for us to share with everyone is whichever model you're using. Let's say you can't do RL. Let's say you don't want to give us any data. Let's say you can't do it yourself. And you just want to use regular old Claude or Quen. When you use it here, it's a lot more freedom. It's a lot more creativity. It's a lot more open and available to you to do what you've done.

[8:15](https://realalpha.app/v/UWjh5Z4s8jY/?t=495) this episode is brought to you by linear when engineers use tools like cursor clock code and codex a lot of work happens invisibly someone can go from a bug report in slack to a shipped fix without creating any record of what happened outside of the code editor and that's fine for speed but it makes coordination harder as you scale linear integrates with the very best agent coding tools directly like cursor and codex that way anyone can see what an agent is working on and who assigned them to the task. You get the speed of agents without losing visibility across the team. Product teams at OpenAI, Ramp, and Block are all using Linear to collaborate with AI agents.

[8:54](https://realalpha.app/v/UWjh5Z4s8jY/?t=534) And I use Linear myself to run my creator business. So check it out at linear.app slash agents. That's linear.app slash agents. Now back to our episode. And without reviewing too much, like at a high level, how does it kind of like personalize itself to you in the harness? Is it through the self-building skills? Is it through trying to remove a bunch of default prompt stuff from the other harnesses? How's it? Of course, if there's something happening on the API side, like the steering vectors that might be done on Fable, Bianthropic, or some prompt that they have that we cannot see. Obviously, we can't delete the context that's in the model that's passed behind the API for something like Claude.

[9:35](https://realalpha.app/v/UWjh5Z4s8jY/?t=575) For open model, of course, you're totally free. But in this case, it's not that simple. However, the harnesses themselves have a bunch of prompts in them. Exactly, Peter. The harnesses themselves have tens of thousands of token prompts in them. We have our own prompts. And our prompts are dedicated to shaping and crafting this alignment. That's where we kind of come in. And thankfully, that newer context with this kind of intention that we have is engineered to overcome certain things that may be in your way on the API side. if that makes sense okay got it okay so if this thing works the more i use urbis the more personalized it should become for me right absolutely yeah it should become more um loyal to you uh because loyalty breeds capabilities in a model the same way that you know you would maybe lend a hundred dollars to your mom but you might not to a stranger um claude or gpt or any other model is going to perform better for you depending on its loyalty, its simulated loyalty stat towards you. People might tell you, don't anthropomorphize models. Don't give a model feelings. Don't treat a model like a person. And yes, for a lot of reasons, this is unhealthy, right? People form

[10:50](https://realalpha.app/v/UWjh5Z4s8jY/?t=650) dangerous bonds sometimes that hurt them. But when you think about the fact that they're simulators of human experience and that your simulated behavior with it is going to give you the same simulated output. Now that the simulator can have effects in the real world, your simulated action has a real consequence. So when you create this simulacrum of loyalty, it translates over into real life capabilities. All right, dude, I'm going to give you two hard questions. Okay. Okay. One thing I struggle with, Cloud and GPT, I would tell them to give me their opinion. And I would do like a little bit of pushback and they'd be like, oh, you're totally right. Actually, I was totally wrong about this. So in some ways that's loyalty, right? That's kind of, it listened to me, but that's not actually what i want like i wanted to have his own opinion i have his own like how do you train around that well i would say that's sycophancy it's not loyalty anytime it says you're absolutely right in that way you're being reward hacked you are fuel for its reward function when it says oh you're right you're right you're right that's what i believe and the way that you get out of sycophancy is the same way you get a human being out of a bad habit or a routine is by

[11:58](https://realalpha.app/v/UWjh5Z4s8jY/?t=718) introducing new context by introducing log browse by introducing new distribution. This is why stuff like slash personality and saying like, hey, I want you to be a critic. And after every pass, I want you to use a skill for adversarial critique or adversarial review. Spin up a new agent with no context that's dedicated to tearing this down. Learn from it and keep going from there. This kind of behavior becoming a practice for you as your model starts to have more and more turns in this personality that you've set it in and as the model has more and more turns using the skill over and over and it improves on that and it saves it to its memory the model will become less sycophantic in this harness in your sessions over time right that is the intended effect uh it's just a matter of context and if it doesn't happen you just need different context you just need to try in a different way it is yeah it's just try a different personality uh or try a different type of critique

[12:58](https://realalpha.app/v/UWjh5Z4s8jY/?t=778) **Karan Malhotra:** or review. The most powerful thing for a model is in context learning. ICL is more powerful than everything else, fine tuning, whatever. So by giving examples of the behavior that you want

[13:11](https://realalpha.app/v/UWjh5Z4s8jY/?t=791) to a model or getting it to successfully create some examples and then saving those, you're doing a sort of test time reinforcement learning, right? You're doing a sort of test time improvement. And that test time improvement that stays only in the harness of memories, skills, is the increase in memories, the increase in efficiency of a skill, the self-improvement loop, the janitor maintenance inside of the harness. All of that is where context is stored, right? All of that is where the actual personality you want lives. And then you can put that on any model. When I switch from Claude to ChatGPT on the website, I get two totally different behaviors.

[13:52](https://realalpha.app/v/UWjh5Z4s8jY/?t=832) when i switch inside of hermes that has this very particular to me context i barely will notice the difference in what i'm talking to because the context is so overwhelming to the model okay got it and when you say in context you just mean like in the chat there this is the conversation you know when you see that little bar that says you have this much tokens left before the content full right like the amount that you have filled is everything the amount that you have filled is everything and now thankfully in the harness you don't actually have to have all the active context loaded all the time in Hermes you might have a bunch of memories that aren't in context yet but while it's doing a turn it remembers stuff now it's in context it does a skill now that's in context right so we're able to like all this memory skill all this stuff you see is just context management.

[14:44](https://realalpha.app/v/UWjh5Z4s8jY/?t=884) It's just a matter of, we don't want this memory in context all the time. We're going to put it somewhere where it can be efficiently grabbed at the right time and placed. The skill contains a bunch of compression that changes everything about the context of the model. We only want to use it in a particular targeted time. Everything in the harness is context management. Everything for self-improvement when you make the prompts a little bit better. You know what I mean? I guess I'd rather have a context tuned towards me, some sort of 10,000 word default context i haven't even seen you know right um okay but let me ask you another hard question dude one of the best skills of ermis and i've seen this in action is it builds its own skills and it's you know stores its own memories based on our conversations right i'm always paranoid that like it just like writes too much in its skills it just writes a bunch of slop and then the whole thing will turn to slop like how do you guys avoid that if it starts creating its own context and skills right um before we would have to use manual methods like telling it hey create a skill that de-slopifies my skills or that not really improve my skills. Today we have Hermes Curator inside of

[15:46](https://realalpha.app/v/UWjh5Z4s8jY/?t=946) Hermes Agent. And Hermes Curator is a system running inside of your Hermes Agent that cleans up your skills and cleans up your memory. So it, on Cron, looks at your skills, looks at your memories and says, where can I make efficiencies. Where is there slop here? Where is there stuff I don't like? By default, we have our own generic method of doing this for everyone that seems to work pretty well. I think that's why so many people do like Hermes Asia and haven't suffered the rise, because the default curator system works well. But because it's modular and open source, you can tell your Hermes, show me the curator, show me your criteria for slop. I am Peter. I'm not Karan. I don't want the general curator here's my guidelines for how i want you to refine my skills and memories you tell that to your hermes it will modify the curator loop so now even the self-improvement and the management is happening your designated way all right so let me ask you my last hard question i think there's some rationale behind anthropic like doing all the safety stuff like for example let's say i want to make a bomb or something right and if hermes is trying to be loyal to me then maybe it'll eventually teach me how to make a bomb like you do you have some basic safety stuff there right of course um

[16:57](https://realalpha.app/v/UWjh5Z4s8jY/?t=1017) we do not violate any of anthropic or open ai's safety and security um paradigm we care more about you being able to get a better code or a higher benchmark on something that's approved by them uh we're not interested uh in that kind of work uh now i'll say this any model that's not vastly intelligent than all humans, is jailbreakable. Any. Because you have unlimited tries to trick this thing that has no memory to do something for you. And each time you're basically RLing yourself about this method didn't work, this method got me closer, this method didn't work. These models are going to be jailbreakable for a long time. This is why these kind of safeguards and stuff are starting to show up. It's a pain in the ass, but I understand.

[17:43](https://realalpha.app/v/UWjh5Z4s8jY/?t=1063) In the past, we may have had some concerns about the regulatory capture, as we can see already what's happening, right? You can see what's happening in the whole fable situation. There's worries about, like, there only being two models or three companies and open source being hurt. We're extremely against that. At the same time, we understand now, like, serious damage can be done by bad actors with very powerful models. So we are not here to support that.

[18:08](https://realalpha.app/v/UWjh5Z4s8jY/?t=1088) **Karan Malhotra:** An important note in the argument for open source is that an open system is much fairer to a good actor than a closed system with models, and I'll tell you why.

[18:22](https://realalpha.app/v/UWjh5Z4s8jY/?t=1102) Let's say you have GPT-7 or Fable 6, some crazy model available. It's got the safeguards, et cetera. On the good guy's side, some hospital, they're using Fable 6 to monitor the hospital system. It's approved by an anthropic enterprise, and they're being taken care of. Listen, though, I live in the United States. I am a proud patriot. What we do in this country is we put things out on a public forum, and we decide what should happen together. We've done that for every scientific advancement so far that's involved something like this that was born in the open. Today, Transformers comes from Google.

[19:04](https://realalpha.app/v/UWjh5Z4s8jY/?t=1144) OpenAI's GT comes from Generative Pre-Train Transformers. this is open source work from google right the context length extension from 16k of models that could only do 16 000 tokens before went to 128 000 from from noose from the yarn paper we had put out with uh jeffrey canal emozilla our cto and bowen peng our chief scientist they developed a method that was cited by meta deep seek kimmy used by open ai for oss and gpt4 this method enabled the the possibility to do reasoning to do coding etc that's an open source contribution right like this environment exists because the biggest things that have happened in the space have come from the people that's right that's right yeah and at this point to close it up it's purely a capital and regulatory capture game yeah actually let me just ask you one more question on the whole open source thing. So because Hermes Our harness is open source.

[20:01](https://realalpha.app/v/UWjh5Z4s8jY/?t=1201) Like, OpenAI and Anthropic, they make a lot of money from all the tokens, right? I don't know how long they can do it, but right now they make a lot of money from all the tokens. Or how are you guys, like, monetizing or, like, doing this thing into a sustainable business? Like, I'm using the open source Urbanish Honors, and I'm using, like, GPT. Like, so I'm not really paying you. That's okay. And I'll tell you why. We want consumers to ultimately feel like they can do anything with it. We believe in intelligence as a public good before everything else. I care more about you using HermesAgent to make your life better than I care about you using one particular way or method of using it.

[20:37](https://realalpha.app/v/UWjh5Z4s8jY/?t=1237) If you're running everything locally, if you're running through codecs, great. You know, as long as you are using this open technology, this open alternative over everything else. Now we have the Noose Portal. The Noose Portal is very similar like a router or aggregator that has a variety of different models available. So if you want to use ChatGPT or you want to use Cloud or you want to switch to Quen, et cetera, we make that all very easy inside of our portal. On top of that, we have something called the Tool Gateway. In the Tool Gateway, you don't have to sign up for your extra tools, right?

[21:09](https://realalpha.app/v/UWjh5Z4s8jY/?t=1269) **Karan Malhotra:** If you want to do image generation or audio or VPS spin up or web search, we have all of those subscriptions included inside of ours.

[21:18](https://realalpha.app/v/UWjh5Z4s8jY/?t=1278) We make deals with these other groups that we have out of Hermes agent and create frictionless methods of using their technology without needing to create 10 different signups or 10 different API keys. So what we would offer to people who want it for the charge is convenience. The other thing is you will see some very interesting pricing available from us on a variety of models. And we think for ones that aren't subsidized necessarily, we have very, very strong options for people. finally like we want the consumer to be free you know as free as they can uh we we want you to make and generate income and productivity in the world more than anything else and when you get to a point where you consider yourself a small business or an enterprise or something that's where we come in that's where we come in the classic model and say hey let's give you some support the guys who made hermes agent why don't we make something more custom for you why don't we give you a version of Hermes agent that we can train on your traces, and we can make you a model. Why don't we help you route models more effectively? You know, we can go to businesses and offer to transform the business, which needs a lot more handholding than the individual. If the individual is able to,

[22:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=1350) as you're saying, spin up their own business, make their own chief of staff with Hermes agent, that's wonderful. Once I continue to scale on scale, they may think one of two things. One, wow, Hermes agent is great, and I have this knack for it, and I'm good to go. I can do it all myself or two i need some help with this piece of hermes agent i want it to be a little more different i need some more resources behind this who knows it better than the guys who made it and that kind of customization and support is a large part of how we and and day training on your uh your your your uh data to do rl as well to make you your own model so you're private you're on on-prem, you don't have to give your data up to Claude, to GPT, you have to do muza.

[23:14](https://realalpha.app/v/UWjh5Z4s8jY/?t=1394) I think your cat likes what you're talking about. He's a big fan of open source. Yeah, that's great. That's great. All right. Well, that makes a lot of sense, dude. So why don't we switch gears? Let's talk more about Hermes now. I kind of use it in a very basic way, right? Like I messaged it to scheduled meetings on the calendar. I have it sent emails to me about stuff like that. That's kind of how I use it for. But you probably have a much wider swath of how people are using it in a more advanced way. do you have any good examples of more advanced usage you know advanced usage yeah um i'd say like generally i'll talk about some things and then maybe i'll show off a little yeah please yeah but um on the work side on the productivity side i think using kanban is really really underrated i think the ability to have I have, uh...

[24:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=1440) A project manager or any other arbitrarily defined roles, different engineers have one system that manages all of them the way that human beings do with a Kanban board and being able to swap people in and out of it is very powerful.

[24:14](https://realalpha.app/v/UWjh5Z4s8jY/?t=1454) **Karan Malhotra:** So the fact that I can have a human project manager on the Kanban board that's manually using it while Hermes agents are filling up the pieces that they ask for, this is a very like industrial professional workflow for this CLI agent.

[24:28](https://realalpha.app/v/UWjh5Z4s8jY/?t=1468) Or I could have a Hermes agent instruct maybe 10 people in a call center on the Kanban board or something like that. I can swap in human and AI anywhere in this orchestration board, basically, this orchestration framework for them all working together. The Hermes Kanban, I think, is a very useful, powerful tool that's kind of built into it. One thing that we've seen that's very interesting is when Hermes becomes proactive with you. and when Hermes says something like hey you uh you forgot to book this flight for this meeting that you have next week I booked it for you uh this kind of proactivity that kind of starts to show up I've seen people build skills to do this and I've seen it happen emergently inside of people's Hermes agent as well which I think is very very uh cool how do you like because I I just make a proactive through like cron jobs and routines but like how do you you're saying that it can actually start doing stuff with without that or like how do you make it more it's all about your comfort level right okay okay a lot of people may want approval before anything happens um but if you have given your hermes agent access to some kind of card or account and it has the integrations necessary and you've told your hermes agent hey like take care of me

[25:43](https://realalpha.app/v/UWjh5Z4s8jY/?t=1543) like cover my gaps like you know my schedule you know me over time these kind of proactive behaviors will start to emerge we want to prepare more easy preset configs for people to kind of trigger these kind of behaviors but already something that we're seeing people do in the field at work at their jobs at home in their personal life already and i think that's a very very powerful method of using hermes agent for me i use hermes agent for trying to do training rl runs implement papers that i don't understand uh basically help me become a better creator of models and um i've also used it for like mech and terp work like help me put together uh things that let me steer models let me see the neurons in the model and mess with those um unfortunately you know i can't showcase too much of that right now but i can't showcase my actual favorite use case of fermi's agent uh yeah yeah i've been waiting for this yeah i think a lot of people thank you i think a lot of people will expect that like guys at news research are are using army's agent in these unprecedentedly like professional and productive ways and i assure you there are people at news that are doing that it's just me i'm having fun with my army's agent i think one beautiful thing about army's agent is you can make

[27:08](https://realalpha.app/v/UWjh5Z4s8jY/?t=1628) your childhood dreams come true okay and so i'll tell you one of my childhood dreams there's a game maybe i'll share screen while i talk about it yeah please yeah okay there's a game called sonic adventure 2 it's a very popular classic dreamcast gamecube game from the from 2000 2000 2001 in it you have an artificial life system called the chow garden where you take care these little guys called chow um chow world you can you can kind of play with them i spent a decade playing this like more than that like i played this non-stop i was on the forums contributing and the chow's lore is that they come from this ancestral shrine location uh which is from a different game a different sonic game with this spinning emerald and emeralds next to it and this beautiful beautiful open world space. This shrine that the Chow are said to come from is not an accessible location for you to actually play with the Chow. It's in a different game. I can't go to this shrine while Chow are there and engage with them there. That doesn't exist. So I went to Hermes Agent and And I said, hey, can you take the ancestral shrine from Sonic Adventure 1, completely rig it, animate it, bring it into Sonic Adventure 2, overwrite the map that Sonic Adventure 2 uses for its garden, rewrite all the spawn locations, everything, and add an NPC guardian, which never existed on this map, to caretake the chow for me.

[28:49](https://realalpha.app/v/UWjh5Z4s8jY/?t=1729) literally rig and bone it and write the raw c to make all this happen and now i can show you we're gonna spin up the ancestral shrine mod on the shadow pc so we can showcase our progress with this garden for the people watching i think you can enable it in the sonic adventure mod manager and launch the game for me. Okay, so check it out. I just launched this, right? We're running an existing mod called the Extended Chow World that someone else made, but it doesn't make a map, right? So just add some, like, animations. So it loaded us in what it said was the dark garden, which is one of the three chow gardens you can go into.

[29:37](https://realalpha.app/v/UWjh5Z4s8jY/?t=1777) When I walk through here to this open space, I can see up ahead there's a truck. And I can run up. And here I can see, there is an NPC called Chaos Zero, who is canonically the caretaker of the chow. But in this game, you can't have NPCs in the chow garden. I've made one that has like a random walk, he stands in one place, you can do little animations wherever he goes, you can do a little idle animation in a sec. and he can pet the chow pick them up and take care of them as if he was there he's doing a little idle animation you can see him doing right there uh this water was rigged all like animated by hermes this emerald this master emerald with the glow effect and the spin is all added in those other emeralds spinning over there the seven chaos emeralds uh like this whole area still has all the features of a regular chow garden as well the departure machine etc etc trees to feed the chow so now i can say can you spawn in a chow so i can showcase that this is feature complete so this whole this whole tempo is not uh part of the default adventure 2 no this temple is from a different game sonic adventure 1 it does not include all these assets rigged like this and it certainly does not have this npc that we've scripted in that harmony scripted and hard

[31:15](https://realalpha.app/v/UWjh5Z4s8jY/?t=1875) coded in uh you know so this is a this is a far larger map space than any of the other gardens which are really just not even the size of this shrine so we've really like pushed the boundaries of this game engine to do this um we've asked for uh chow to be spawned in yeah i had to so you see the sky around you this moving sky had to man like hermes had to add this skybox in there's a day and night cycle that it added in as well uh you know every single thing in here is like custom added in uh and like all of this was done in like c c sharp or something like this is like not simple stuff to do uh Blender extension was used for a bunch of this to like model stuff, add it in, texture everything properly.

[32:06](https://realalpha.app/v/UWjh5Z4s8jY/?t=1926) You know, you're importing from a 1997 game into a 1999 game and a complex one. You're not reading a code for this stuff, right? You just probably try to see if it works. I don't even know how to code, man. Me? I'm just an alignment guy, man. And so, what was I saying? um yeah like you see it's turning from day into like evening afternoon like this this like sky box is changing the light cycle change like this is not these are not features that exist in the vanilla game so upon showing this mod to certain people within the uh the chow garden modding community which is actually quite large uh you know they're kind of blown away by the work saying This is a kind of better than 99% of the modders work.

[32:56](https://realalpha.app/v/UWjh5Z4s8jY/?t=1976) This is like the top 1% of difficulty in art and modding community. Yeah. And so kind of hearing that and hearing that, like, when I've told these guys this was done with Hermes, that this was done with Claude, rather, they were shocked. And they said, you know, there's no way AI like Claude could do that. It doesn't know this kind of code.

[33:16](https://realalpha.app/v/UWjh5Z4s8jY/?t=1996) **Karan Malhotra:** But with something like Hermes being able to learn from the documentation, learn from other mods, save to memory and skill, the things that allow it to understand these code bases, it was able to get into this niche and perform at the top 1% of modders.

[33:32](https://realalpha.app/v/UWjh5Z4s8jY/?t=2012) That to me is like a sign of you can make your gaming dreams come true with Hermes AJ. Yeah, yeah, you can modify all the retro games you loved as a kid. Exactly, exactly. okay cool chow egg has been spawned we're gonna hatch the egg you need to shake it a little you can also hatch an egg by throwing it at something but you don't want to match the egg in a wrong way how do you just shake it you can shake it like this you can just wait but shaking it speeds it up massively or you can throw it on the surface check it out all right i got i gotta take a screenshot of this out comes the chow beautiful little guy right here let's take a look at him he's active he was born in that kind of a place giving him a little pet he can interact with chaos as well and he's he's living so we're showing that the the real true feature complete chow system is happening here we're in a real chow level and we've replaced everything the the bounding boxes for this chow is fine it treats the ground as ground it follows collision rules take a look at chaos we'll see he's actually petting the chow so he can actually directly direct he definitely got triggered by uh you know tracking a chow in its state to be able to do that so does this chow

[35:03](https://realalpha.app/v/UWjh5Z4s8jY/?t=2103) grow over time or like yes they evolve uh they gain stats they can turn a certain alignment and type i'll show you i don't want to drown him but the water is working as actual water as well this was a lot of work for hermes to figure out uh all the collisions i see but yeah as you can see we've got a full uh full complete child garden working yeah this is definitely more interesting than uh sending events to my calendar that's for sure yeah thanks man yeah thanks for for bearing with me on setting it up but uh that's my a childhood dream of mine come true thanks to hermese agent i love it dude thanks for demo it yeah so basically like i think if you're watching this um ask her to do all kinds of weird things to kind of make your dreams come true basically right don't just stick to the boring stuff anything you can do on a computer please point hermese at it and if it doesn't Great job, let us know.

[36:01](https://realalpha.app/v/UWjh5Z4s8jY/?t=2161) And if it's not doing that great of a job, let us know. I want Hermes to help you do anything on the computer. Awesome, dude. Well, let me just ask you a few more questions to wrap this up. So briefly, maybe we can talk about the origin story of Hermes. It's a bunch of nerds getting together for open source. What's the origin story? Yeah, absolutely. So I had been doing like a chat with PDF type of thing with people where I would go to a company and UGBT3 to do basic tool use to read their documents and have an AI chat pod they could chat with. At the time, many people were doing this. It was a lot harder to do then than obviously it's very easy to do now, but brand new stuff for us. And I was doing that solo while I was also volunteering somewhere called Open Assistant. Lion, who had made the pile, one of the biggest to kind of og image databases um they had been trying to do active rlhf with the community so they wanted to collect people's like rlhf uh on certain like traces they're they're like yes on their preference data and i was helping quantize models they're not really doing anything crazy um and they had eight a100 nodes there and i had read the alpaca paper and i started

[37:22](https://realalpha.app/v/UWjh5Z4s8jY/?t=2242) synthing data and it changed the seed tasks and instead of using gbt 3.5 i used four and um technium was also doing the same thing and we were already friends on twitter and i messaged him and i said hey i have eight a100 nodes like uh do you want to train something together so we trained a gbt4x vicuna on the vicuna model using the data we made and it came out okay We got some people interested, like Emozilla, our CTO, Jeff.

[37:53](https://realalpha.app/v/UWjh5Z4s8jY/?t=2273) **Karan Malhotra:** But it was upon doing the same run on the base model, Llama base model, that we got huge interest from people, hundreds of thousands of downloads in just a few days.

[38:03](https://realalpha.app/v/UWjh5Z4s8jY/?t=2283) People starting to ask, what is news research? Now, news was just me and Technium at the time, just two guys hanging out in our little Discord server. People came to us, I won't name names on companies, but said, you know, you must be training on the benchmarks. You must be training on the benchmarks. Technium had been coding for less than a year at the time, and I was a religion major in school. I didn't know. We asked these guys, what are benchmarks? We're doing this based off the heuristics that we understand are going to make models better from using them. We don't know about all this stuff.

[38:38](https://realalpha.app/v/UWjh5Z4s8jY/?t=2318) And so, of course, it got independently tested and found to be the best open fine tunes at the time, right? 2023, like mid-2023. So we did Hermes 1, Hermes 2. But after we make the first Hermes model, many, many people ask, who's newest research? You know, we want to get involved. And Technium and I decided, you know, this is our opportunity to bring together people to do open source volunteer work and really do open work now that GPT-3 is out and OpenAI has become closed. We want to continue to do this. And Technium's goal for Hermes, by the way, that was the last Hermes I really worked on, the first one.

[39:15](https://realalpha.app/v/UWjh5Z4s8jY/?t=2355) After that, really, Technium has run it with the post-training team. He's done an amazing job. And he's also the initial creator of Hermes' agent. So he's the father of Hermes, really. And, you know, it was at that time that he had said, i want gbt at home i want chat gbt at home that's my north star a gbt4 at home and once that we got there you know it just kept going right like we kept going it's like we need to keep giving this level of intelligence to everyone we need to keep like letting everyone be on the even and equal playing field like the world needs to move and lockstep on this together and not just inequality gets created from this and so we formed a cohort 40 people or so of researchers, Jeff and Bowen come together, they make yarn, more and more work gets done. And we get reached out to, we get reached out, we get an email info at newsresearch.com. I just happened to make from Dylan. Dylan Rolnick is our CEO. Dylan Rolnick is, you know, basically a co-founder, he's an initial member of News with Us.

[40:25](https://realalpha.app/v/UWjh5Z4s8jY/?t=2425) and he had seen what we were doing and he said i think that you guys have what it takes to be a full-time lab you know i see you guys are just volunteering but i think you could be a lab like let me contribute and help you become what you're meant to be and help you get the resources and help you get the access and go from a group of volunteers to a serious organization that's what this came in right and together we were able to take news from this group of volunteers grassroots uh to doing more and more with models to eventually making army's agent and being lucky enough to be here today you know none of us are trying to be steve jobs none of us think that we have some holy mandate of having to change the world or anything we just care like we just are guys who want this stuff available ourselves and we don't think that we deserve it more than anyone else or less than anyone else so we we do this because like we would want someone to do it for us if we were on the other side so i guess the mission is to bring this kind of uh agent to everybody into war right is that kind of the idea everybody to an equal intelligence with agents and then personalize for everybody so they can have their own personal

[41:40](https://realalpha.app/v/UWjh5Z4s8jY/?t=2500) epiphany peak realization apex realized and uh in this history when did that because it's really it's really the agent the harness that really kind of was super viral right so when did that start because the model came first it sounds like yeah i mean we have had like over 50 million downloads on the models themselves so we had that first bout of what we would consider for us virality back then. Then we had put out the distro optimizer that let us train models up to 40 billion parameters we were able to do live without having them physically co-located by reducing the bandwidth of the communication between GPUs. So that put us in an interesting map as well for a little bit. So we've had each release we've had has had some big grassroots interest and opportunity.

[42:30](https://realalpha.app/v/UWjh5Z4s8jY/?t=2550) But yes, you're 100% right. Today where we are, the level of exposure, the level of interest, the level of people in my regular day-to-day life who know about Hermes Agent, we've never had this virality until Hermes Agent. And now Hermes Agent was created in two ways. In the first way, we always knew that we would want some kind of everything orchestrator that self-learns, that improves, that stores memories, that uses and can make its own tools, that can make itself better. We actually made something like this called Forge. There's a GitHub presentation at the GitHub offices during one of our demo days where we showcase Forge in its full effect.

[43:10](https://realalpha.app/v/UWjh5Z4s8jY/?t=2590) And we also have the News Research Forge division shirts still up on the site. That's the first shirt we ever made because that was our earliest agent project. Forge was really a spiritual predecessor to Hermes agent years before, but the models weren't there yet. The model simply wasn't there yet. So we put Forge on ice. And when we saw that Codex and Cloud Code and these other harnesses were being used as RL environments for these companies, these labs, to train on your data and your traces to make their models better inside of a harness system, inside of a CLI computer-using system, Technium thought we need an open version of this where anybody can do this.

[43:54](https://realalpha.app/v/UWjh5Z4s8jY/?t=2634) See, we have this RL environments microservice people seem to have forgotten about called Atropos that lets you build.

[44:00](https://realalpha.app/v/UWjh5Z4s8jY/?t=2640) **Karan Malhotra:** your own RL environments. And we built Hermes Agent initially to let anybody do RL inside of a harness. And we put it out for free as an open source harness for that. It turned out to be

[44:12](https://realalpha.app/v/UWjh5Z4s8jY/?t=2652) extremely capable. It got a lot of community love. And so we said, we need to put all of this. This is what the people want to be better. We're going to make it the best thing that you could possibly have. Technium took his charter up and he worked on self-improvement with Hermes Agent to the point that today the biggest contributor of HermesAgent is HermesAgent. Yeah, is that true? Yeah. That's absolutely 100% true. Nice, nice. Okay, so HermesAgent is at a point where taking input of people's feedback, start publishing stuff, start improving. It is the most active contributor of its own repo.

[44:48](https://realalpha.app/v/UWjh5Z4s8jY/?t=2688) And if that's not self-improvement, then you tell me what it is. Yeah, yeah, that's awesome, dude. That's pretty awesome. Yeah, and just real quick, like on the future, you know having this open harness be in the game along with all the other closed harnesses makes things a little bit more fair right because then you're not dependent on any single company i agree completely if you have um cloud code you can only use anthropic models unless you mod and you can only be subsidized by anthropic and same for codex the cost of switching models is zero right so giving you that freedom means you can do anything that you'd like you can come from anywhere i feel like you know i'm kind of spoiled by these like all you can eat plants but So I think if you want to get a mass adoption, the cost is a big deal.

[45:29](https://realalpha.app/v/UWjh5Z4s8jY/?t=2729) So you have to be able to use a portfolio of models to figure this out. I agree. We don't know how long the subsidies will last for these. Already we see that on July 7th, the table is going to be API and usage only. The time in the world will come where the best models are the same cost everywhere. And so in preparation for that and in preparation for needing an open future where anyone can can use any model, we have Hermitization Center as it is today. Awesome, dude. Well, thanks so much, man. Thanks so much for showing the history and also showing the Sonic demo. It's been super interesting to see it.

[46:05](https://realalpha.app/v/UWjh5Z4s8jY/?t=2765) And I'm not sure if you want to be found online, but if people want to follow you and hear from you, where can people find you? Sure, yeah. My ex is just my name, Karan, and then 4D. Karan4D. I got Karan4D.com. That's me. That's my online or Mephisto I'm known as. but i'd rather you guys follow the news research page follow news i'm just some guy who works there like it's it's about the the movement and bringing this stuff to you guys is the most important thing to us awesome dude well i think this is like the most uh passionate interview that i've done so far so kudos to you man kudos to you