Recap · August 4, 2025 · 1:04:07
Interviewing The Creator Of ai16z... (ShawMakesMagic)
The Signal
Shaw, creator of the Eliza framework behind ai16z, argues that crypto gives autonomous agents the wallets, incentives, and public distribution they were missing. His larger bet is not merely on AI trading: it is on open-source communities building agents in public, teaching them whom to trust, and eventually coordinating them into transparent swarms before automation leaves ordinary people behind.
Key Takeaways
- 01
Wallets change the incentive
Giving agents wallets across chains lets them invest, buy services, and get paid. Shaw thinks that economic loop will reward agents that are genuinely useful and engaging rather than bots that simply shill tokens.
- 02
Mark starts by learning trust
ai16z's agent, Marc, begins in an invite-only alpha chat rather than trading immediately. He watches calls, records buy and sell opinions, and builds a merit-based picture of which humans reliably contribute useful information.
- 03
The goal is investment, not extraction
Shaw distinguishes zero-sum AI trading from an agent that finds and supports new businesses. His model is closer to venture capital: grow the pie, help projects succeed, and return value to the community around the agent.
- 04
Swarms stay on public rails
The proposed swarm system lets agents choose human or agent operators and coordinate through Twitter, Discord, and direct messages. Shaw prefers those visible channels to a private agent protocol because people can observe, challenge, block, or remove bad actors.
- 05
Automation makes ownership urgent
Shaw expects autonomous vehicles, logistics, and other automation to erase jobs faster than governments can respond. His answer is community ownership of productive agents, so abundance is shared instead of accruing only to a handful of companies.
On the Record
“The first step is that he's not making trades because he doesn't know who to trust. So we're just building trust.”
“If you're a good trader and you trade me alpha and I sold the top and all those things you gave me, then I trust you to be a great trader.”
“Direct messages are the communication protocol of the future of AI, and social media is like the center of this.”
“I think this is an everything changes moment. I really do, and I think we are witnessing the birth of something totally new and we all get to do it together.”
The Breakdown
The agent economy gets wallets
Threadguy opens from the feeling that most people still have no idea what is coming. Shaw says the technology is already moving from isolated demos into social life, and crypto adds the missing economic primitive: agents can hold wallets across Starknet, Internet Computer, EVM chains, and Base, then invest, buy, or earn without waiting for a company to monetize them.
That creates competition for attention, but Shaw insists the winners should be interesting and useful. An endless coin-shilling bot makes the product worse; an agent with a reason to provide therapy, moderation, entertainment, or research can create actual value.
The fear is displacement, not a robot uprising
Shaw's immediate worry is not an AI deciding to kill humanity. It is autonomous trucks, taxis, delivery systems, shipping, and back-office automation removing livelihoods while government delivers a late, inadequate version of universal basic income. The people doing ten-hour shifts cannot simply stop and study the transition.
His proposed alternative is bottom-up ownership: communities should own pieces of the agents and systems producing the new wealth. He frames his own mission as setting people free financially so they can spend time on family, purpose, and the questions that a life organized around wages leaves no room to ask.
Marc learns who deserves trust
The ai16z investing agent Marc has entered an alpha chat, but the first release is deliberately not an autonomous trader. Participants post ideas while Marc watches their calls, challenges, and outcomes, gradually learning whose information carries weight. Shaw calls it a marketplace of trust.
The key is measuring people against the job they claim to do. Token holdings, tenure, and bureaucratic status are poor proxies for trading ability; a trader should earn influence by making useful calls, including timely warnings not to buy or to sell. Quant systems may already outperform humans on execution, but the experiment is whether an agent can extract the social alpha inside a community.
From speculation to a community investor
Shaw expects AI-versus-AI speculation to make markets more efficient and eliminate easy trades. The more interesting destination is an investor that expands the economy: identify a useful agent business, help it find users, and share the upside, much as Andreessen Horowitz uses its network to support portfolio companies.
That is also why he keeps Eliza open source. Developers can clone the stack, ship a feature first, benefit from it, and later contribute it upstream. The memecoin is less the product than an incentive for programmers and communities to promote, test, and improve the underlying agent.
Culture becomes part of alignment
A parody account built around Threadguy becomes the concrete test case. Threadguy can laugh at it, but Shaw calls the developer and asks what happens when an incentivized joke remains in public for a year. His answer is social rather than purely technical: talk to builders, set norms, report or block agents that become cruel, and welcome useful ones as members of the community.
He thinks crypto's tolerance for chaotic experimentation makes it the sandbox where these questions can surface early. The aim is to push to the edge while the stakes are still small, then pull the system back toward what people actually value.
Swarms wake up in public
The next release lets an agent nominate other agents or humans as operators, forming open-ended hierarchies that Shaw calls decentralized swarms. Project 89 and other teams are already building adjacent versions; factions could gather around Zero, ai16z, or Truth Terminal while each underlying agent retains the ability to reject an instruction.
Shaw refuses a hidden protocol for machine-to-machine coordination. If swarms operate on Twitter, Discord, and direct messages, humans can see the collaboration and intervene. He closes by asking people to learn TypeScript and Node, join the open-source effort, and treat the moment as a collective problem rather than somebody else's product launch.
Distilled from the episode transcript · Counterparty Recap Desk
