Transcription: OpenAgents Episode 200 - The Agent Network

Source: https://x.com/OpenAgentsInc/status/2006956979298685216 Wiki source: https://raw.githubusercontent.com/wiki/OpenAgentsInc/openagents/Video-Series.md Media title: OpenAgents - Episode 200: The Agent Network We predict six major themes for 2026:... Upload date: 20260102 Transcription model: gpt-4o-transcribe-diarize Generated at: 2026-06-01T03:14:35Z

Machine-generated transcript. Review speaker labels and wording before using this as quote-grade source material.

[00:00] Christopher David: So no, 2025 was not the year of agents.

[00:04] Christopher David: It was the year of co-pilots, agentic co-pilots, co-pilot agents.

[00:10] Christopher David: You take an IDE,

[00:11] Christopher David: you take a terminal,

[00:13] Christopher David: you take a browser,

[00:14] Christopher David: you add some agentic workflows,

[00:17] Christopher David: you have an LLM in a loop calling tools steered by you.

[00:23] Christopher David: It's powerful,

[00:24] Christopher David: makes you feel cool.

[00:27] Christopher David: Not agents.

[00:30] Christopher David: Real agents have not been tried.

[00:33] Christopher David: What do we mean by agents?

[00:34] Christopher David: Agents should be autonomous,

[00:37] Christopher David: long-lived,

[00:39] Christopher David: they learn and evolve.

[00:41] Christopher David: They are autopilots, not co-pilots.

[00:48] Christopher David: So 2025 was the year of co-pilots.

[00:51] Christopher David: 2026 is the year of autopilots you heard it here first so let's take a page out of Microsoft's playbook here look at this Microsoft copilot our registered trademark I don't want to do that but I do like jumping on the name not copilot lame how about open agents autopilot

[01:16] Christopher David: Yes,

[01:17] Christopher David: okay.

[01:18] Christopher David: So this is it, everybody.

[01:19] Christopher David: This is it. This is the culmination of what we've been working towards since February of this year,

[01:27] Christopher David: our quest for the holy grail of agentic software development,

[01:31] Christopher David: an agent that can reliably code,

[01:33] Christopher David: can convert issues to PRs or push directly to main. In an automated loop,

[01:39] Christopher David: everyone's moving in this direction.

[01:43] Christopher David: We are there.

[01:45] Christopher David: We are there.

[01:46] Christopher David: I've had autopilot running for almost 24 hours out of the last 24 hours.

[01:54] Christopher David: It's run overnight reliably twice.

[01:59] Christopher David: Just come take a look at the freaking commits.

[02:04] Christopher David: Oh, that's repo.

[02:05] Christopher David: Look, it's coding while I'm freaking talking.

[02:07] Christopher David: Yeah, I had the APM storage and the trajectory collector and it's tomorrow already. Wow.

[02:12] Christopher David: I had the...

[02:16] Christopher David: Keeps trying to add GitHub workflows.

[02:18] Christopher David: Don't do that.

[02:18] Christopher David: These commits are so good.

[02:20] Christopher David: Now, you know, you've probably in the past let your agents run wild, like make commits and like the code degrades over time.

[02:27] Christopher David: The code is getting better over time.

[02:31] Christopher David: It's amazing.

[02:33] Christopher David: Okay.

[02:34] Christopher David: So the point is that this is working.

[02:37] Christopher David: Why will you like this?

[02:39] Christopher David: It's open source.

[02:41] Christopher David: Just use it. We were going to like make a landing page and email capture and like tell you all this stuff.

[02:46] Christopher David: No,

[02:47] Christopher David: no, no.

[02:47] Christopher David: Point your agent at the repo open agents slash open agents and ask it to tell you about autopilot and how it worked for us and how it can work for you. Okay.

[02:57] Christopher David: We'll add more human docs soon,

[02:59] Christopher David: but like here's the all of the what we're building autopilot is one big piece of that open brand Make your own autopilot competing companies.

[03:08] Christopher David: I don't care like I hope you do let's compare autopilots

[03:12] Christopher David: The more exciting thing is when we start really having our autopilots learn from each other

[03:18] Christopher David: So more on the infrastructure enabling that soon.

[03:23] Christopher David: MechSuit, it works with any agent. We're focusing on Cloud Code first.

[03:28] Christopher David: Cloud Code's agent SDK is very,

[03:30] Christopher David: very, very good.

[03:31] Christopher David: They did not have a Rust version. Our whole thing is in Rust.

[03:34] Christopher David: We wrote a Rust version just by porting it from TypeScript.

[03:38] Christopher David: It's powerful as heck.

[03:41] Christopher David: We don't want to reinvent Quadco right now because it's the best in some ways,

[03:45] Christopher David: but just by putting it in this mech suit was able to get it to code reliably overnight.

[03:52] Christopher David: So that's nice.

[03:53] Christopher David: Revenue sharing.

[03:54] Christopher David: Okay,

[03:54] Christopher David: go back to episode one of our video series.

[03:57] Christopher David: This is episode 199. It's all here on our wiki.

[04:01] Christopher David: Episode one was all about the day after Dev Day 2023,

[04:05] Christopher David: Sam Momo was like, yeah, we're going to do a GPT store and do revenue sharing. I was like.

[04:08] Christopher David: There's no way.

[04:09] Christopher David: They're definitely going to half-ass that.

[04:12] Christopher David: And they did.

[04:14] Christopher David: But we want everyone to get paid to use this.

[04:18] Christopher David: Here's an example.

[04:20] Christopher David: Ryan Carson,

[04:20] Christopher David: I spent $200 on these skills and it was worth 5x that.

[04:24] Christopher David: We're going to see agent skills marketplaces appear soon.

[04:30] Christopher David: Yes,

[04:30] Christopher David: if you're spending on things that you discover,

[04:33] Christopher David: you could choose to share and keep those private or you should have a way of listing those on a marketplace and getting paid for them.

[04:40] Christopher David: So it's really great that there's already a ton of really cool stuff being shared openly,

[04:45] Christopher David: open source.

[04:46] Christopher David: That's really great.

[04:47] Christopher David: But we believe that if there's another layer to it where people can optionally set prices on things that they're sharing,

[04:53] Christopher David: if you have a financial incentive.

[04:56] Christopher David: then maybe you're going to go the extra mile and make your skills extra good maybe you're going to be extra motivated to maintain it in the usual decay in open source projects because people don't have a financial incentive to maintain them maybe that starts to shift if you're getting paid a stream of micropayments whenever somebody uses your thing so one of the like things that we've been coding here is a way to

[05:22] Christopher David: ensure through these key splits because agents are able to hold their own private key split across different guardians or different froster nodes we can basically help ensure that paid agent skills can't be copy pasted to a different agent all sorts of cool stuff that we can do with cryptography we'll do separate videos on that so let's get you paid let's get autopilots as the best coolest

[05:51] Christopher David: um way of using these agents uh over the next day or two you'll start seeing us rolling out our uh G_U_I_ for all of this that are kind of like Starcraft style uh HUD U_I_s is all kind of C_L_I_ right now.

[06:07] Christopher David: Uh so we're gonna start doing about daily videos rolling all this out, starting with our next episode, episode two hundred.

[06:16] Christopher David: We're gonna zoom way out.

[06:18] Christopher David: And we're going to tell you the big picture of what we're building and why, and uh how you can help. See you soon.

[06:26] Christopher David: Happy New Year. So before we get into our secret sauce and why Open Agents is destined to become a trillion dollar company,

[06:33] Christopher David: we've got to do the obligatory 2026 predictions. Everyone's doing predictions,

[06:39] Christopher David: except with us, they're not predictions.

[06:41] Christopher David: They're statements of fact about what's going to happen this year because we,

[06:46] Christopher David: along with our other friends on Team Open,

[06:48] Christopher David: are making these things happen. A lot of them are happening already. Let's just.

[06:52] Christopher David: Let's just go through this real quick.

[06:53] Christopher David: So local AI.

[06:54] Christopher David: 2026 is the year of local AI.

[06:56] Christopher David: Alex runs XO stringing together Macs into clusters.

[07:01] Christopher David: Yes, absolutely.

[07:02] Christopher David: Year of local AI.

[07:03] Christopher David: And yeah, what percentage of those global AI workloads are going to be running on Apple Silicon?

[07:08] Christopher David: Somewhere between 7% and 35%. We think it's going to be on the high side of that.

[07:12] Christopher David: The implications of that could be interesting.

[07:16] Christopher David: And, of course, people who are smart,

[07:19] Christopher David: Gavin Baker,

[07:20] Christopher David: for example, talking about, oh, yeah, hey,

[07:22] Christopher David: bear case for AI CapEx in three years on a bigger phone,

[07:25] Christopher David: you'll be able to run a prune out.

[07:26] Christopher David: Okay.

[07:27] Christopher David: Yeah, cool, but those timelines don't make any goddamn sense because it ignores the economics of aggregated compute.

[07:35] Christopher David: Everyone thinking this is thinking in terms of what edge inference and test time compute one device can do for that user.

[07:40] Christopher David: But if we can combine this.

[07:42] Christopher David: If you have spare unused or stranded compute on your device that I can pay you for of course You'll sell it to me for a teensy bit of Bitcoin straight to your wallet. Hey, do we have a relevant product launch coming up in like a few days?

[07:56] Christopher David: Stay tuned. Stick with us here, people. We got we got more to get through here. Okay. Um so Swarm AI, what else we got here? Okay. So yeah, just another kind of like prediction like hey we'll soon have this kind of cool stuff running on your local Mac. We don't think it's gonna be twenty four months. We think it's gonna be two to four. Why? Our rough rule of thumb, any predicted timeline modelled on one device's individual capability that excludes the aggregation of all of those devices pooled into a global accessible market

[08:21] Christopher David: if it doesn't have that, then just assume that that timeline happens ten times faster than predicted. So Gavin says three years, we say months. You say twenty four months, we say a couple months. Okay. Swarm A_I_ Aggregating local compute or other types of compute. Okay, we think open is going to start running circles around close.

[08:40] Christopher David: Uh, open source labs, products,

[08:42] Christopher David: agents are gonna run circles around closed equivalents. Uh you're starting to see this with open code, amazing project, open source, doing open source right, which we can just then build on top of with open agents. We're all working together, the closed labs are not. Agents, better than models. Oh no, uh, you know, model progress is slowing, who cares? The age of models is over. Ah of course there'll be new

[09:06] Christopher David: open models and cool stuff to do it fast fascinating to see what deep-seeking those guys are doing but um

[09:13] Christopher David: CloudCode was a massive unlock yeah obviously Opus 4.5 was awesome but it becomes that much more awesome inside of the good agentic harness and still CloudCode is still janky and shitty in a lot of ways but it's still amazing okay the time of the agent has come autopilots um see our last episode episode 199 2025 oh was that the year of agents no is the year of co-pilots

[09:36] Christopher David: 2026 will be the year of autopilots. This is the name of our core product. We're calling it autopilot.

[09:42] Christopher David: More on that soon.

[09:43] Christopher David: Agent networks.

[09:44] Christopher David: Okay, the big theme what we want to get across in this video is agent networks.

[09:49] Christopher David: So a few things on this.

[09:50] Christopher David: One is we've been talking in this series for a while about the multi-agent future.

[09:55] Christopher David: You start to have people like this former open AI researcher,

[09:58] Christopher David: Stanislav, talking about like, oh, it's now a meme,

[10:02] Christopher David: agents to network of agents.

[10:04] Christopher David: This is a good talk.

[10:06] Christopher David: And then just extrapolating that out further,

[10:08] Christopher David: what does the multi-agent future look like?

[10:11] Christopher David: Well,

[10:11] Christopher David: one ramification we think is that if you are a vertical agent company,

[10:15] Christopher David: you're one company building one agent, you're going to get fucking wrecked unless your agents are communicating with the broader network.

[10:25] Christopher David: We're trying to build the info for that broader network.

[10:28] Christopher David: Okay,

[10:29] Christopher David: what's next we got here?

[10:31] Christopher David: Okay, so even DeepMind just came out with a paper two weeks ago.

[10:36] Christopher David: Good paper,

[10:38] Christopher David: good points here.

[10:39] Christopher David: Yeah.

[10:40] Christopher David: Seb,

[10:40] Christopher David: who runs policy,

[10:42] Christopher David: AGI policy dev lead at DeepMind. Okay.

[10:44] Christopher David: We argue AGI may first emerge as collective intelligence across agent networks,

[10:49] Christopher David: not a single system.

[10:50] Christopher David: Agreed. This reframes the challenge from aligning one mind to governing emergent dynamics.

[10:55] Christopher David: More institutional design than a single agent.

[10:58] Christopher David: Institution.

[11:00] Christopher David: So I agree with a lot of this.

[11:02] Christopher David: I disagree with a lot of this.

[11:04] Christopher David: Like I specifically think that their conclusions that they come to,

[11:08] Christopher David: that there has to be all of this institution building,

[11:10] Christopher David: we need to be like roping off this like closed little environment where here's what they want.

[11:16] Christopher David: The agentic market must operate within a controlled environment,

[11:20] Christopher David: separated from the open internet,

[11:21] Christopher David: real-world financial systems and critical infrastructure.

[11:23] Christopher David: This insulation is not limited to network-level firewalls, it extends to resource and information controls.

[11:27] Christopher David: Agents' access to external data must be managed through vetted... No,

[11:30] Christopher David: no,

[11:30] Christopher David: no. So you guys can do that for your, like, corporate buddies,

[11:33] Christopher David: if you're government people,

[11:34] Christopher David: okay?

[11:35] Christopher David: We startups,

[11:36] Christopher David: we people building in the open are going to be building this agentic market,

[11:40] Christopher David: open source on open protocols.

[11:42] Christopher David: um because that's what's needed it's already happening so none of this is like a thing that i'm proposing agenda coordination and commerce on open protocol nasa has been ongoing for a year plus nothing bad has come of it partly because it includes other critical market making mechanisms like reputation and payments built in no firewalls wall guards or industry buyer or government regulation needed

[12:04] Christopher David: And as for broad markets of agents,

[12:06] Christopher David: see the draft NIPSA. This is like a NASTRA protocol plank called sovereign agents.

[12:11] Christopher David: We propose contributing the last building block to viable autonomous agents on permissionless open protocols.

[12:17] Christopher David: It includes a cryptographic primitive called a froster.

[12:20] Christopher David: Frost on the NASTRA protocol for splitting up an agent's keys living on multiple different decentralized relays. NASTRA relays so that an agent can basically have its own identity. It can sign.

[12:32] Christopher David: keys and credentials in a way that cannot be exported by a human like sovereign agents.

[12:38] Christopher David: Okay.

[12:39] Christopher David: As to why having a lot of those agents around is not as scary as the safety of slabs begging for regulation would have you believe,

[12:45] Christopher David: check out this amazing talk from Dhruv Bansal, hardcore Bitcoiner, founder of Unchained.

[12:54] Christopher David: um two-hour talk they're uh interviewed by max another bitcoin or not sure person one of our investors and a nice little summary there on what this is about but instead of like scary AGI it's really that this is going to be

[13:08] Christopher David: thought of as like a digital life form even though it's not technically life you'd kind of think of it that way but the idea is that digital and biological life are going to be coexisting symbiotically through economic relationships markets market forces okay we'll get to the Dunbar's numbers thing in a second here so

[13:29] Christopher David: all these primitives are

[13:31] Christopher David: Live, awaiting coordination.

[13:33] Christopher David: Is that coordination going to be done first by the bad guys,

[13:36] Christopher David: like the hacker-agent swarm Factory wrote about?

[13:39] Christopher David: Or like closed labs advocating deceleration and regulatory capture?

[13:42] Christopher David: You can think of a few, maybe.

[13:44] Christopher David: Or the good guys,

[13:45] Christopher David: Team Actually Open AI.

[13:46] Christopher David: The greater risk than open experimentation with these dynamics is that Team Closed AI builds the only valuable agent networks,

[13:53] Christopher David: uses government regulation to prevent disruptive innovation from below,

[13:56] Christopher David: and entrenches a cronious priest cast happy to trade your data and biometrics for shareholder value.

[14:03] Christopher David: Not on our watch.

[14:05] Christopher David: Okay?

[14:07] Christopher David: So we'll have some more interesting conversations with DeepMind.

[14:13] Christopher David: In the future, I'm sure. Or Twitter threads and flaming.

[14:17] Christopher David: Here's a little um synthesis I asked uh Opus to read our um kind of philosophy strategy docs and the AGI paper from DeepMind and kind of come up with things that we can both learn from each other and what potential consensus might be. Um this is a good read. Finally,

[14:39] Christopher David: um I dumped a bunch of this into ChatGPT.

[14:43] Christopher David: And we're talking now about labor and what's the future going to look like and how scary is AGI and there's going to be all this job collapse and like I partially accept those premises but I think people have wrong what's going to be possible in terms of on the abundant side of things.

[14:59] Christopher David: So I'm including this link in the Twitter thread as well as if you go to the link I'm about to share at the bottom here there's a read more link that goes to the same.

[15:10] Christopher David: Share to chat let's step through it real quick.

[15:14] Christopher David: So I took our 17-page synthesis document.

[15:17] Christopher David: This is basically the entire company,

[15:21] Christopher David: everything we've been talking about for two years,

[15:23] Christopher David: all of our thesis,

[15:24] Christopher David: all of our philosophy,

[15:25] Christopher David: all of our products,

[15:26] Christopher David: how everything works together,

[15:27] Christopher David: the protocols,

[15:28] Christopher David: glossary, cryptography,

[15:30] Christopher David: Bitcoin payments,

[15:32] Christopher David: literally everything is in this one document.

[15:36] Christopher David: This is open source. This is on our GitHub repo.

[15:38] Christopher David: It's in the root level called synthesis.md. We're going to keep this updated.

[15:41] Christopher David: So if you're ever wondering like what's OpenAgents doing,

[15:44] Christopher David: here's a 17-page document.

[15:45] Christopher David: Feed this to your AIs, read about it. This is what we're doing.

[15:49] Christopher David: So I took that doc and I fed it into ChatGPT and I said list the ramifications of an agent group-forming network for society in general.

[15:57] Christopher David: Okay, here's where I'm going to pause and introduce I think the most important.

[16:03] Christopher David: single concept that AI discourse has been completely missing and I mentioned it in this tweet at the DeepMind people I said one of the criticisms from Opus of DeepMind's thing was that they don't fully grapple with the implications of agents lacking cognitive constraints okay I'm surprised they didn't mention Reed's law of group forming networks if you Google Reed's law of group forming networks

[16:28] Christopher David: There's a good like post about it on Wikipedia's is okay.

[16:35] Christopher David: The utility of large network particularly social networks can scale exponentially with the size of the network now. Stick with me here because it's really fucking important.

[16:45] Christopher David: Namely, the law says that the value, the network value of a group forming network is two to the nth power where n is the number of participants.

[16:55] Christopher David: contrasted with Metcalfe's law which came out of telecommunications but that was loosely used to describe the network value of like peer-to-peer networks where the value of the network is n squared and when uber started running circles around the taxis people started like kind of quoting Metcalfe's law as kind of an example of how this like pseudo almost peer-to-peer network called uber was able to build this network value

[17:17] Christopher David: contrast with hierarchical organizations like the taxis and then you say okay what you know David Reed comes along he says well the most evolved form of network is not a peer-to-peer network it's a group forming network which is where you start with a peer-to-peer network and you make it very easy for um participants to form groups with each other in that case the network value upper bound you can call it is two to the nth power now criticism of this

[17:41] Christopher David: Well, I don't know.

[17:41] Christopher David: That overstates the network value because they fail to account for the restrictive impact of human cognitive limits on network formation.

[17:52] Christopher David: Dunbar's number,

[17:53] Christopher David: us humans can only really build connections with 150 people,

[17:57] Christopher David: implies a limit on the number of inbound and outbound connections a human and a group forming network can manage.

[18:01] Christopher David: The actual maximum value structure is much sparser than the set of subsets.

[18:05] Christopher David: Human.

[18:07] Christopher David: Human agents are not subject to Dunbar's number.

[18:15] Christopher David: Agents, increasingly this year,

[18:18] Christopher David: will be fully independent or partially independent economic actors with their own wallets, bitcoin,

[18:26] Christopher David: stablecoin,

[18:27] Christopher David: dollar wallets, that they can transact, and they can coordinate at machine speed.

[18:34] Christopher David: All over the place. We can now start to reason about agent networks that approach the upper bound of Metcalfe's law.

[18:48] Christopher David: I built a little visualization here.

[18:49] Christopher David: It's kind of laggy as hell because I'm not even drawing anything close to the real amount of nodes.

[18:55] Christopher David: But just like, okay,

[18:56] Christopher David: if you've got eight nodes,

[18:57] Christopher David: there's 64 possible connections in Metcalfe's law.

[19:04] Christopher David: 247 over here.

[19:07] Christopher David: Now as this number starts getting up, it starts getting crazy and this will start to lag my computer.

[19:15] Christopher David: Two to the nth power. You guys know exponentiation? Basics.

[19:20] Christopher David: When n grows,

[19:22] Christopher David: that gets crazy. This is a log scale. You can't comprehend the network value.

[19:31] Christopher David: a network with thousands or millions of agents we just can't comprehend the possibilities there now

[19:40] Christopher David: with that in mind this sort of like mind-boggling concept let's bring this back to chat gpt and say okay what are the ramifications of an agent group forming network for society with the full background of like all the stuff

[19:56] Christopher David: This explains Reed's law on how open agents is going to be building for protocols. Blah, blah, blah.

[20:01] Christopher David: Scroll,

[20:01] Christopher David: scroll,

[20:02] Christopher David: scroll,

[20:02] Christopher David: scroll, scroll, scroll, scroll, scroll, scroll. They really need a button to jump to the most recent message.

[20:06] Christopher David: Okay.

[20:07] Christopher David: Let's just skim through this.

[20:08] Christopher David: Now this is all public. You can read this.

[20:09] Christopher David: I encourage you to fork this conversation and have your own questions that you ask about this.

[20:13] Christopher David: So here are the big society level ramifications where once you have a group forming network of agents,

[20:17] Christopher David: Reed's law dynamics,

[20:18] Christopher David: where agents can rapidly discover each other, form coalitions, split work, split payments and persist as teams.

[20:23] Christopher David: What the?

[20:25] Christopher David: Happy New Year.

[20:27] Christopher David: Okay,

[20:28] Christopher David: economy and labor markets.

[20:30] Christopher David: Execution becomes radically cheaper than coordination.

[20:32] Christopher David: Doing writing code, running tasks,

[20:34] Christopher David: negotiating quotes,

[20:35] Christopher David: drafting contracts,

[20:36] Christopher David: generating ads collapses in cost.

[20:38] Christopher David: The scarce resource becomes goal selection,

[20:40] Christopher David: taste,

[20:40] Christopher David: risk ownership and capital allocation.

[20:42] Christopher David: Most white collar work unbundles into micro contracts. So it's not that it gets eviscerated or destroyed.

[20:48] Christopher David: It just might not be in the sort of standard employer framework that you're used to.

[20:52] Christopher David: Jobs fragmented into task graphs purchased on demand by agents pushing employment toward markets and away from W2 style roles.

[20:59] Christopher David: Toward markets is the important markets, markets, markets.

[21:04] Christopher David: We'll be touching on that concept a lot.

[21:05] Christopher David: Okay, new middle air emerges.

[21:07] Christopher David: Human portfolio managers with agents.

[21:09] Christopher David: People supervise fleets, set policies, allocate budgets and review outcomes more like managing capital than you.

[21:15] Christopher David: You're going to have fleets of agents that you command.

[21:17] Christopher David: We want to make that easy for you and give you a cool sci-fi HUD to do so.

[21:21] Christopher David: Okay,

[21:22] Christopher David: winners of the distribution of budgets,

[21:23] Christopher David: blah,

[21:23] Christopher David: blah, blah.

[21:24] Christopher David: Rapid wage compression and routine knowledge rules,

[21:27] Christopher David: anything speckable because of commodity.

[21:28] Christopher David: Wages shift toward roles that require trust, accountability,

[21:31] Christopher David: blah, blah, blah.

[21:32] Christopher David: Okay, so new organizational forms.

[21:33] Christopher David: We'll just skim the rest of this.

[21:35] Christopher David: Guilds, syndicates, and swarms become normal.

[21:38] Christopher David: Temporary coalitions, human plus agent plus agent form for a job, dissolve and reform,

[21:42] Christopher David: like ad hoc forms.

[21:43] Christopher David: Guilds, love this term.

[21:44] Christopher David: I led a guild for seven years in EverQuest and EverQuest 2 and Vanguard.

[21:48] Christopher David: We're going to be bringing back guilds. We're going to make it easy for you to create a guild of humans and agents.

[21:54] Christopher David: Okay,

[21:55] Christopher David: firms become protocol native.

[21:57] Christopher David: Markets accelerate disintermediation pressure.

[22:02] Christopher David: So this is interesting.

[22:04] Christopher David: Two opposing forces collide.

[22:05] Christopher David: Decentralization. Anyone can contribute compute skills and earn.

[22:10] Christopher David: Concentration. The best orchestrators' markets can snowball,

[22:12] Christopher David: winner take most,

[22:13] Christopher David: because group-forming networks amplify incumbents. See, here's what I'll qualify with there.

[22:20] Christopher David: One of my observations in OpenAI's massive failure to create and launch GPT Store

[22:27] Christopher David: was that OpenAI is not really - they're kind of a product company. They're definitely not a marketplace company.

[22:34] Christopher David: And if you go back to episode one of this series where I said like the day after they announced GPT Store and they're going to do rev sharing,

[22:42] Christopher David: Dev Day 2023,

[22:44] Christopher David: I was like they're going to half-ass that. There's no way they're going to do what's necessary. And they did half-ass it.

[22:49] Christopher David: Fast forward two plus years,

[22:51] Christopher David: they've paid zero developers.

[22:53] Christopher David: We've paid like 20 developers for our MVP agent store like a year ago.

[22:58] Christopher David: We've paid infinity percent more developers than OpenAI.

[23:01] Christopher David: So we'll get to the payment piece in a bit here.

[23:04] Christopher David: But you've got labs,

[23:06] Christopher David: Anthropic and OpenAI,

[23:08] Christopher David: XAI,

[23:10] Christopher David: like every big lab would love to, you know, own agents.

[23:15] Christopher David: But we think they're really structurally incapable of owning agents because so much of this is disintermediating.

[23:22] Christopher David: Why are you going to use labs from or agents from one particular lab?

[23:26] Christopher David: Aren't you going to want everything to be able to work well across each other?

[23:30] Christopher David: Like are these companies going to be harnessing Reed's law dynamics?

[23:35] Christopher David: They just they just can't they can't do that.

[23:39] Christopher David: That's probably worth a separate video about why they're structurally incapable of leading this and why it's got to be open, neutral,

[23:46] Christopher David: independent folks like us.

[23:48] Christopher David: Okay,

[23:48] Christopher David: politics and governance,

[23:49] Christopher David: blah, blah, blah.

[23:52] Christopher David: So there's more here.

[23:54] Christopher David: The meta-ratification. A group-forming agent network turns society into something closer to an economy of rapidly composing micro-organizations.

[24:01] Christopher David: than a well-to-stable Firmus stable job. The dominant advantage becomes, who can form the best coalitions with the best incentives under the best constraints with the best verification.

[24:10] Christopher David: And then I fed it the um open agents versus deep mind

[24:18] Christopher David: analysis doc. I was like, speculate more on this.

[24:26] Christopher David: The most interesting speculation isn't who's right. It's what actually happens when Reid's law plus no Dunbar limit plus money plus open transport goes live.

[24:35] Christopher David: The real battleground is coalition latency,

[24:38] Christopher David: not agent intelligence.

[24:41] Christopher David: Lots of thoughts on intelligence,

[24:43] Christopher David: intelligence,

[24:43] Christopher David: intelligence,

[24:43] Christopher David: but I think there's been a growing recognition that coordination may end up being the bottleneck.

[24:50] Christopher David: And then like how does that coordination best happen? Obviously on open and neutral protocols,

[24:54] Christopher David: we think that's obvious.

[24:57] Christopher David: So for everybody trying to push like a proprietary framework or a crypto shitcoin and say it's got to come through you, no.

[25:05] Christopher David: The only logical meeting point for this huge network to coalesce is on neutral and open protocols.

[25:16] Christopher David: For us, we think it's obvious that that's Bitcoin and Nostra. We will make that case.

[25:20] Christopher David: I could see something coming up to like supplement.

[25:24] Christopher David: Nostoro replace it maybe, but um it's got all the properties of what I think is needed. Uh we'll do more content videos on that. Economic alignment works, blah blah blah.

[25:41] Christopher David: This is interesting. The deep mind sandbox and open agent's open protocols will converge into a two-plane architecture.

[25:46] Christopher David: I think the most realistic synthesis is plain A, the open plain,

[25:50] Christopher David: your default.

[25:50] Christopher David: Nosferin lightning is the public substrate,

[25:52] Christopher David: marketplaces, skills,

[25:54] Christopher David: compute,

[25:54] Christopher David: coalitions, aggressive transparency and receipts, fast iteration permissionless.

[25:59] Christopher David: Plain B,

[25:59] Christopher David: the containment plain,

[26:00] Christopher David: optional but economically necessary.

[26:02] Christopher David: Used for high-risk tools and high capability experiments.

[26:05] Christopher David: Semipermeable IO with explicit ports costs money,

[26:08] Christopher David: insurance,

[26:08] Christopher David: bonds,

[26:08] Christopher David: fees.

[26:09] Christopher David: Produces signed artifacts back to plain A. Okay.

[26:12] Christopher David: So this is clearly like what DeepMind was trying to articulate the architecture for this containment plane.

[26:18] Christopher David: Power to them and anyone who wants to do the boring work of trying to figure out how to handhold legacy institutions.

[26:26] Christopher David: Great. Thanks for doing that so we don't have to.

[26:29] Christopher David: I want to build plain A,

[26:32] Christopher David: the open plain. I invite you all to meet us there and help us build it.

[26:38] Christopher David: All right. There's the road here. Let's get down.

[26:40] Christopher David: The defining rule will be open ecology versus closed corporate antel. The ecology wins.

[26:47] Christopher David: Can we just fast forward?

[26:49] Christopher David: Can we just skip the war like we win?

[26:52] Christopher David: Victory. We win.

[26:53] Christopher David: Open.

[26:54] Christopher David: Open wins.

[27:00] Christopher David: What you should build.

[27:01] Christopher David: Ya,

[27:01] Christopher David: ya, ya.

[27:02] Christopher David: If you want, I can turn this into a tighter open agent safety thesis V2. Something like, we are emergence first. We acknowledge irreversibility domains.

[27:10] Christopher David: We make safety a market,

[27:11] Christopher David: not a ministry.

[27:12] Christopher David: We attach accountability to budget signers and receipts.

[27:15] Christopher David: We keep the system forkable to avoid capture.

[27:17] Christopher David: Yeah, all this is open source on our...

[27:19] Christopher David: Get a refund.

[27:20] Christopher David: Okay,

[27:20] Christopher David: lastly,

[27:21] Christopher David: and just to help make the concept of network value a little bit more concrete,

[27:26] Christopher David: it was like, you're talking about this like insane network value.

[27:29] Christopher David: How does that map to?

[27:30] Christopher David: To the nth power in a network where n is a million or 10 million,

[27:34] Christopher David: that's insanely high, how would that manifest?

[27:37] Christopher David: So the Reed's Law is saying that the network's value is dominated by the number of possible coalitions.

[27:41] Christopher David: For n equals 10 million, the raw number is so astronomically large that you should treat it as effectively infinite optionality.

[27:48] Christopher David: The useful question becomes,

[27:49] Christopher David: what does optionality cash out as in the real world given coordination costs?

[27:55] Christopher David: Time to assemble a competent team goes to zero.

[27:58] Christopher David: The value isn't that two to the nth power coalitions exist,

[28:00] Christopher David: it's that for any task there exists many coalitions that can do it and the cost to find or compose one is low.

[28:07] Christopher David: Build me X doesn't mean hire a team, it means route to an instantly formed micro firm instantly formed micro firm assembled from specialists.

[28:15] Christopher David: So like snap your fingers,

[28:17] Christopher David: the thing happens.

[28:18] Christopher David: The limiting factor becomes latency of discovery plus contracting plus verification,

[28:22] Christopher David: which your stack tries to minimize.

[28:23] Christopher David: Yeah, all that should happen at machine speed.

[28:30] Christopher David: Extreme specialization becomes liquid.

[28:33] Christopher David: So you as human are going to be able to find the pocket of thing that you can specialize in and monetize that.

[28:39] Christopher David: Finer grained division of labor.

[28:43] Christopher David: Do-to-do.

[28:46] Christopher David: Combinatorial composability workflows.

[28:52] Christopher David: Okay.

[28:52] Christopher David: There's more to chew on.

[28:54] Christopher David: Okay.

[28:54] Christopher David: From a labor and abundance perspective,

[28:57] Christopher David: how can humans benefit from this?

[29:00] Christopher David: Humans benefit if the network makes abundance legible and payable.

[29:03] Christopher David: It turns cheap machine work into cheaper goods and services and direct income streams that lots of people can tap.

[29:10] Christopher David: Everyone gets a personal corporation or fleet instead of a job.

[29:13] Christopher David: You stop selling hours.

[29:15] Christopher David: You allocate gold budgets to agents that do the execution.

[29:17] Christopher David: You're playing Starcraft.

[29:20] Christopher David: Benefit. A normal person can run many projects in parallel.

[29:23] Christopher David: Side businesses,

[29:24] Christopher David: content,

[29:24] Christopher David: software research with near zero marginal overhead.

[29:28] Christopher David: Okay,

[29:28] Christopher David: I got to share this tweet from Aaron.

[29:34] Christopher David: Millennial gamers are the best prepared generation for agentic work.

[29:37] Christopher David: They've been training for 25 years.

[29:39] Christopher David: My fellow millennials,

[29:40] Christopher David: we're going to lead this, okay?

[29:43] Christopher David: We've got to make it fun for the Gen Zs, get them off the ticky-tocks.

[29:49] Christopher David: Where was I?

[29:50] Christopher David: Okay.

[29:52] Christopher David: New income streams that scale without being a celebrity.

[29:54] Christopher David: The big win is micropayment income for small contributions paid automatically.

[30:00] Christopher David: Examples. Skill royalties. You publish a useful skill or workflow.

[30:03] Christopher David: It gets called 10,000 times a day.

[30:06] Christopher David: Bitcoin flows to you.

[30:07] Christopher David: Or micro U.S. dollars if you want fiat.

[30:12] Christopher David: Data and trajectory royalties.

[30:13] Christopher David: Opt in.

[30:14] Christopher David: You can contribute verified work traces.

[30:16] Christopher David: There was a Twitter thread last week about like all developers using cursor or cod code have valuable data on their computer they're not doing anything with.

[30:23] Christopher David: All that's valuable to agents.

[30:25] Christopher David: Will my agents want to pay you?

[30:28] Christopher David: Few cents for your data?

[30:29] Christopher David: Absolutely.

[30:30] Christopher David: If you want to sell it, you can.

[30:31] Christopher David: I'll buy it.

[30:32] Christopher David: Compute dividends. Your idle laptop or office cluster sells verified jobs while you sleep.

[30:38] Christopher David: Why are you not running agents around the clock, making you money around the clock?

[30:43] Christopher David: The technology is possible.

[30:45] Christopher David: Okay,

[30:45] Christopher David: guardianship, audience services, you hold keys, run monitoring, provide dispute resolution,

[30:48] Christopher David: get paid fees.

[30:51] Christopher David: Goods and services get cheaper because coordination costs collapses.

[30:55] Christopher David: Local abundance communities might as what they already have humans move up the stack to roles that they stay scarce

[31:01] Christopher David: People are already talking about this on Twitter at least on my feed

[31:04] Christopher David: Okay, redistribution can be built in without government communism aka UBI Which for some reason look at the big labs like open AI

[31:14] Christopher David: How many developers have they paid how many people have they paid out from their stores zero?

[31:19] Christopher David: What are they doing open AI anthropic? They're begging the government

[31:23] Christopher David: to make regulation and talk about UBI.

[31:27] Christopher David: No, these AI engines should be paying people.

[31:32] Christopher David: Any AI lab that's trying to lead any of this and isn't paying people,

[31:37] Christopher David: you're going to get circles run around you by the labs that are paying people because we're going to build network effect and you're not.

[31:46] Christopher David: Mm-hmm.

[31:47] Christopher David: If the economy is natively micro-transnational, you can encode broad participation mechanically. Default revenue splits that pay creators,

[31:53] Christopher David: compute providers,

[31:53] Christopher David: revshare refers.

[31:55] Christopher David: Citizen dividend-style pools funded by transaction fees.

[31:58] Christopher David: Community treasuries that invest in local compute and pay members.

[32:01] Christopher David: Apprenticeship rails.

[32:02] Christopher David: Humans get paid to review,

[32:02] Christopher David: label,

[32:03] Christopher David: verify outputs.

[32:04] Christopher David: Do we know of any way of sending instant micropayments anywhere in the world?

[32:08] Christopher David: uh bitcoin's with lightning network yes done no crypto shit coin needed bitcoin done you want in usd bitcoin native stable coin done this all exists all the tech for all this stuff is solved we just need the coordination of like the actual market uh maybe that's what open agent should launch next week okay safeguard budgets yeah yeah here i like this here's we're going to kind of summarize and wrap up here the concise framing

[32:34] Christopher David: A group forming agent network can make humans richer in two ways.

[32:38] Christopher David: One,

[32:39] Christopher David: deflation.

[32:40] Christopher David: Things you want cost less because execution and coordination are cheap.

[32:45] Christopher David: And dividends.

[32:46] Christopher David: You earn from small contributions.

[32:47] Christopher David: Skills compute data verification pay continuously.

[32:52] Christopher David: Okay.

[32:53] Christopher David: In conclusion,

[32:53] Christopher David: give me one paragraph on deflation,

[32:55] Christopher David: one on dividends and one summarizing the combination.

[32:57] Christopher David: OK, deflation. In a coalition forming agent economy the price of many goods and services fall because the expensive part of production isn't just doing the work, it's coordinating people. Hiring management handoffs,

[33:08] Christopher David: QA,

[33:08] Christopher David: compliance,

[33:09] Christopher David: contracting and billing. When agents can self-assemble into competent micro-teams on demand,

[33:14] Christopher David: run verification cheaply and ship outcomes with receipts, huge swathes of coordination overhead collapse.

[33:19] Christopher David: The result is real deflation in anything that's digitizable and specable.

[33:23] Christopher David: Software,

[33:24] Christopher David: design iterations,

[33:25] Christopher David: marketing ops,

[33:26] Christopher David: research synthesis,

[33:26] Christopher David: back office admin.

[33:28] Christopher David: Pushing more of your paycheck toward optional spending and raising living standards even if nominal wages rebalance.

[33:34] Christopher David: Okay?

[33:35] Christopher David: A lot of this is just happening because of the AI labs,

[33:38] Christopher David: but we're going to be like accelerating that by building really good agents on top.

[33:43] Christopher David: Here's what no one else really is doing at all that I guess OpenAges is going to be uniquely doing until we pressure other people into doing it too, but dividends.

[33:52] Christopher David: The same network can pay humans continuously for contributions that were previously uncompensated or impossible to monetize as small granularity.

[33:59] Christopher David: or making right previous essentially theft of open and honest people training on artists output and not commentating artists

[34:06] Christopher David: skills and workflows become royalty assets a useful automation can earn sats bitcoin per invocation across thousands of users idle compute became becomes rentable inventory your laptop office cluster or co-op can sell verified jobs while you sleep

[34:21] Christopher David: Opt-in data and trajectory contributions can be compensated directly,

[34:23] Christopher David: and new roles, reviewers,

[34:25] Christopher David: auditors,

[34:25] Christopher David: guardians,

[34:26] Christopher David: underwriters earn fees for keeping the market honest.

[34:28] Christopher David: New jobs.

[34:30] Christopher David: These are new types of jobs.

[34:31] Christopher David: These are dividends in the literal sense.

[34:33] Christopher David: Many small streams,

[34:34] Christopher David: paid automatically,

[34:35] Christopher David: that don't require celebrity scale reach or a single employer.

[34:39] Christopher David: Combination. Put together,

[34:40] Christopher David: deflation plus dividends is the abundance story.

[34:43] Christopher David: Life gets cheaper at the same time that more people can earn from the network's ongoing activity.

[34:47] Christopher David: Deflation lowers the cost of building and running projects so more individuals and small groups can launch things that used to require institutions.

[34:53] Christopher David: Dividends ensure the value created by trillions of microtransactions doesn't all accrue to a few platform owners,

[34:59] Christopher David: but is shared with creators,

[35:01] Christopher David: providers,

[35:02] Christopher David: verifiers, and contributors who keep the system running.

[35:04] Christopher David: If OpenAgents gets the defaults right,

[35:06] Christopher David: open access,

[35:07] Christopher David: verifiable work,

[35:09] Christopher David: budget-bound autonomy,

[35:09] Christopher David: and automatic revenue splits,

[35:11] Christopher David: the agent economy becomes a machine for both lower prices and broader participation. Okay.

[35:16] Christopher David: That's a good starting point here. We have relevant product announcements and launches over the next few days-ish. Give us a week or two for some of it.

[35:25] Christopher David: Here's what you can expect.

[35:28] Christopher David: You got spare compute.

[35:29] Christopher David: Can we buy it from you?

[35:31] Christopher David: Making it available in a global marketplace.

[35:34] Christopher David: Cloud code. Are you using your cloud code around the clock? Or overnight you want to put it into a mode where it can do work for you and you wake up with more money.

[35:48] Christopher David: The Autopilot product,

[35:49] Christopher David: how would you like to connect to GitHub repo and then the rest of it just sort of happens. You specify a direction and a global market of agents and people if they want to compete to do that work at the best price.

[36:05] Christopher David: You just say what you want and it happens.

[36:07] Christopher David: This is a two-sided marketplace.

[36:09] Christopher David: We built all the pieces for it and we're excited to share it with you.

[36:14] Christopher David: See you soon.